Secondary Logo

Journal Logo

Original Research

Effects of Fatigue on Bilateral Ground Reaction Force Asymmetries during the Squat Exercise

Hodges, Stephanie J1; Patrick, Ryan J1; Reiser, Raoul F II1,2

Author Information
Journal of Strength and Conditioning Research: November 2011 - Volume 25 - Issue 11 - p 3107-3117
doi: 10.1519/JSC.0b013e318212de7b
  • Free



Functional asymmetries are side-to-side differences in kinematics, kinetics, or both that occur during task performance. Functional asymmetries may arise from a multitude of sources, including anthropometry, neurology, and strength (24). It has been suggested that those who are functionally symmetric have improved physical performance (19,31) and reduced risk for injury (19). These reports are consistent with the discovery that asymmetry and poor alignment lead to increased energy expenditure and decreased muscular efficiency, which ultimately leads to increased stress and fatigue (2). In general, fatigue has been associated with an increased risk for injury (21). For instance, neuromuscular fatigue may alter motor control and coordination, which have been thought to increase the risk for low back pain (29). It has been demonstrated that central fatigue is a key player in anterior cruciate ligament (ACL) injury risk (17). Because most people display at least small levels of functional asymmetry (6,15,16,23,24), if asymmetries increase with fatigue, they may then partially explain this increase in injury risk that occurs with fatigue.

The free-weight barbell back squat exercise is a cornerstone of many resistance training programs, functionally mimicking many activities of daily living and sport. Examinations of functional asymmetries during the squat exercise in healthy individuals have revealed bilateral differences in the net joint moments in both recreationally trained men (6) and long jumpers (11), and in ground reaction forces in healthy collegiate women softball players (19). Significant loading differences between legs imply that a risk of differential development may unfold from uneven workout stimuli, necessitating specific strength training for each leg. Sports practice and competition may perpetuate strength imbalances, because asymmetric movements are often inherent (19). Fitness enthusiasts who lift weights on their own may also be of concern through improper lifting techniques, poor program design, and a host of individual factors.

One critical limitation of the existing knowledge base on the squat exercise is that assessment has occurred in the unfatigued state. It is unclear if this is a good representation of the state the activity is performed in during the majority of a workout. In the limited investigations that have been performed related to asymmetries and fatigue, bilateral asymmetries appear to increase. Bilateral asymmetries were found to increase with fatigue in a sustained isometric maximal voluntary elbow flexion contraction with right-handed oarsmen (20). Asymmetries also increased with fatigue in a swim bench exercise for those who consistently breathed to the same side of the body (22). It is, however, uncertain if these upper extremity protocols on already asymmetric populations translate to the squat. If they do, previously observed low levels of asymmetries might be more important than currently thought levels. No studies examining fatigue relative to asymmetries have been conducted on groups without a predisposition to being asymmetric. However, if the level of effort is assessed (which could be associated with fatigue because maximum force declines with fatigue), it has been found that asymmetries remain constant (6) or decline slightly (3). Furthermore, if speed of movement is assessed (which could be associated with fatigue because movement slows down with fatigue), it has been found that slower movements are more symmetric (12).

Therefore, the goal of this investigation was to examine functional bilateral asymmetries of the lower extremities during the squat exercise in the context of how they may change during the course of an individual set, and across multiple sets, as measured by vertical ground reaction forces (GRFvs) in healthy trained individuals with no predisposition for asymmetric performance. The GRFv was chosen because it is the largest loading component on each leg, though technique variations from individual to individual limit conclusions relative to specific effects at the muscle and joint level. Comparisons of bilateral anthropometric measures with GRFv were also performed to ensure there was no predisposition relative to leg length. Finally, because many of the consequences of asymmetrical loading result from consistent expression of the asymmetry, an examination of test–retest reliability was conducted. The results of this investigation will serve as a foundation for future studies investigating the role of functional asymmetries during the free-weight barbell back squat exercise and other less constrained movements.


Experimental Approach to the Problem

To accomplish the goals of this investigation, a cross-sectional research design was implemented, consisting of 2–3 visits per subject, in which data were collected during each visit. The assessment of GRFv asymmetries during performance of the free-weight barbell back squat was part of a larger project that also included assessment of vertical jumps and horizontal hops. The first visit involved anthropometric measurements, familiarizations to the testing protocol, obtaining maximum single-leg hop distance on each leg, and determination of 8 repetition maximum (8RM) for the free-weight barbell back squat. During the second visit, subjects completed 5 maximal vertical jumps followed by three 50-cm single-leg horizontal hops on each leg. Five sets of 8 repetitions (5 × 8) of the squat with a 3-minute rest between each set were completed at 90% of the subject's predetermined 8RM. The jump tasks performed before squatting were performed again immediately after completing the squat protocol. In both visits, the subjects were allowed ample warm-up and stretching opportunities before familiarization with the protocol and data collection. Approximately half of the subjects returned for a third visit to evaluate the repeatability of the protocol. The focus of this analysis was on the 5 × 8 squat protocol.


Nine men and 8 women (age = 22.3 ± 2.5 years; height = 170 ± 9 cm; body mass = 73.4 ± 13.8 kg; squat 8RM = 113 ± 35% body mass) participated after providing university-approved written informed consent. All subjects were identified as recreational athletes who had been involved in resistance training programs whereby the squat had been incorporated for the past 8 weeks (at minimum) and were currently involved in weekly sporting activities that used the jump (basketball, volleyball, racquetball, etc.). Exclusion criteria included self-reported pain, injury, and soreness at the time of each visit. Any injuries must have healed at least 4 weeks before participation and those with a history of back or lower limb pain, major previous surgery, bone, joint, or muscular disorder, history of neurologic or orthopedic dysfunction, or pain that would limit the ability to perform the squat or jumps correctly were excluded. Any subjects with a known reason to perform the activities asymmetrically were also excluded (e.g., clinically diagnosed limb length discrepancy, ACL reconstruction, bilateral corrective devices, highly trained in an asymmetric skill). Finally, women who were pregnant at the time of investigation were excluded. Subjects were instructed to abstain from performing heavy exercise of the lower limbs and back for 48 hours before each visit. Caffeine consumption was limited to normal daily intake, and no other ergogenic aids were allowed for consumption during the day before the visit. All testing on individual subjects was conducted at approximately the same time of the day to avoid any diurnal variations in performance.


Visit 1 consisted of anthropometric measures: thigh length (inguinal crease to proximal patella, foot placed on a chair with hip and knee at 90°), lower leg length (knee crease to lateral malleolus when standing), foot length (heel to end of longest toe), thigh circumference (identified at the midpoint obtained from the thigh length, foot placed on a chair with hip and knee at 90°), calf circumference (identified at the widest part of the calf and measured in a relaxed standing position with the heel on the ground, toes up, and weight shifted to the opposite leg), and total supine leg length (anterior superior iliac spine to medial malleolus [5]) for each leg. The average of 2 measurements per subject was compiled for reporting and assessment purposes. Standing height and body weight were also recorded.

A 5-minute warm-up on an upright or recumbent stationary cycle ergometer was required for all subjects before additional physical exercise. Stretching was allowed and determined by each individual subject with the provision that a stretch performed on one side of the body must be performed on the contralateral side. Three to 5 submaximal vertical jumps with hands on the hips were practiced, followed by determination of a subject's single-leg maximum hop distance, which included approximately 5–6 hops per leg. Proper free-weight barbell back squat form for each subject was evaluated relative to National Strength & Conditioning Association (NSCA) guidelines (4) with technique flaws corrected. Subjects were allowed to adjust and hold the bar on their back and shoulders in a manner that felt comfortable and balanced to them. Subjects stood facing a blank wall ∼3 m in front of them without foot stagger. Eye gaze and head tilt were not controlled. A lifting belt was made available for those in which it was a part of their normal squat routine. In addition to a standard 22-kg Olympic-sized barbell, weight plates, and spring collars, a “cage”-type squat rack was used with safety pins adjusted to the bottom range of barbell travel (Figure 1). Weight plates were verified by mass measurement on a scale.

Figure 1
Figure 1:
Squat cage positioned over force platforms with visible safety pins and depth bar. Note: Spotters normally stood at each end of the bar.

Determination of each subjects' 8RM was established and defined as the weight that could be lifted no more than 8 times with acceptable form. To determine the 8RM, after a squat specific warm-up of 1–2 light sets, subjects began with a weight assigned by the investigator based on the individual's current weight training program. Subjects completed up to 10 repetitions of that weight and were then asked to stop. After a 3-minute rest, the weight was increased by 2.26 kg. This procedure continued until an 8RM was determined. If the initial load was too heavy, the load was reduced accordingly for the next attempt. On average, 2.2 ± 1.0 sets were needed to determine their RM, with 4 sets the most. A 2.54-cm diameter plastic depth bar, attached to vertical uprights with Velcro, was positioned behind the subject at the lowest point of their squat to ensure each repetition was performed to the same level (Figure 1). A successful repetition required the subject to touch the depth bar before initiating the up phase of the squat. Two spotters were used during all lift attempts, one on each end of the bar. Verbal encouragement was provided for all exercises.

Visit 2 was completed within 2 weeks after visit 1 (except for 1 subject who was affected by temporary data capture equipment failure, completing the second visit 43 days after the first visit). All subjects were surveyed at the start of visit 2 to ensure no change in health status since visit 1. Ground reaction forces were sampled at 1,000 Hz during all tasks of the Jump Protocol and Fatiguing Squat Protocol (described below) with 2 commercial force-measuring platforms (model 4060-10, Bertec Corp., Columbus, OH, USA) mounted side by side and flush to the surrounding floor (Figure 1). It was previously determined that these force platforms measured within 0.1% of each other within the range of loads of this study.

Jump Protocol

After a warm-up similar to that of their first visit, subjects stood with feet parallel (1 foot completely on each force platform with feet approximately shoulder width, no stagger) and were instructed to look straight ahead, with their head erect and hands on their hips. Performance of 1 approved trial consisting of 5 continuous maximum effort vertical jumps without rest then commenced. After this, subjects hopped (1 leg take off, same leg landing) from 1 force platform to another at a distance of 50 cm. Three approved trials on each leg, alternating right and left legs, were performed. These jumps were then completed again, immediately after the fatiguing squat protocol.

Fatiguing Squat Protocol

The same equipment, setup, and squat specific warm-up were used as described in the first visit. Research assistants loaded the squat bar with 90% of the subject's previously determined 8RM. Subjects then performed 5 sets of 8 repetitions with a 3-minute rest between each set. A brief pause (∼1 second) between repetitions was inserted via verbal coaching to ensure each repetition was completed before the next was initiated. Otherwise, subjects performed repetitions at their own pace.

To verify the reliability of this experimental design, 8 subjects returned to the laboratory for a third visit to establish day-to-day repeatability of the anthropometric measures, single-leg maximum jumps, and symmetry performance of the Jump and Fatiguing Squat Protocol. A minimum of 48 hours from the second visit was required before the third visit, with no more than 2 weeks separating the 2 (expect for 1 subject who was affected by temporary data capture equipment failure, completing the third visit 59 days after the second visit). Again, no change in health status was verified at the start of the visit.

Statistical Analyses

Vertical ground reaction forces were low-pass Butterworth (fourth order, recursive) filtered at 10 Hz to remove noise. The average and peak instantaneous GRFv under each foot for the first 2 and last 2 repetitions of each squat set were analyzed (Figure 2). The start of a repetition was set at the time when the Total GRFv (sum of right and left) dropped below body weight plus bar weight. The end of a repetition was the time when the Total GRFv returned to body weight plus bar weight upon completion of hip and knee extension. Individual foot GRFvs were normalized by the total GRFv and converted to a percentage. In this way, a perfectly symmetric repetition would yield a value of 50% under each foot for both the average and peak GRFv. The GRFv asymmetry level was then calculated as the left GRFv % minus the right GRFv % (%L–R), In this way, a value of 2%, for example, would indicate 51% on the left foot and 49% on the right foot, and −2% would indicate 49% on the left foot and 51% on the right. The peak instantaneous asymmetry calculation was based on each foot's instantaneous maximum, not the maximum of the Total GRFv. The results from the first 2 repetitions of each set were averaged for both the average and peak instantaneous GRFv asymmetries, as were the last 2, so that a representative GRFv asymmetry was generated for the beginning (R1 and R2) and end (R7 and R8) of each set.

Figure 2
Figure 2:
Exemplar total vertical ground reaction force (GRFv; top medium gray line), left GRFv (bottom black line), and right GRFv (bottom light gray line) for an 8-repetition set. Vertical lines identify the boundaries of repetitions 1, 2, 7, and 8.

Means and SDs for anthropometric and biomechanical data were compiled. To examine the relationship between initial and final repetitions across sets, 2x 5 repeated-measures multivariate analysis of variance (MANOVAs) were conducted on time to complete each repetition and the level of GRFv asymmetry (%L-R). Because the goal of the investigation was to assess change in asymmetry level and those who started very symmetrically in R1 and R2 of set 1 would not be expected to change (which was verified throughout examination of the results), MANOVAs were also run on a subset of the subjects with an initial asymmetry. The +/−1.7% asymmetry level was chosen relatively arbitrarily. It was a level that excluded highly symmetric individuals yet left a reasonably sized sample set to be evaluated (n = 12 for both average and peak instantaneous asymmetries). Because some subjects began with greater GRFv on the left side and some on the right, the first assessment was on the absolute level of GRFv asymmetry. Within this analysis, all GRFv asymmetries were assigned a positive value, except if subjects switched sides from their initial value. In this case, it was assigned a negative value. Besides assessing the whole group in the manner described above, an additional analysis was performed on the subset of those that started with more GRFv on their left foot to start R1 and R2 and another analysis on the subset of those with more GRFv on their right foot to start R1 and R2. These MANOVAs were performed on both the average and peak instantaneous asymmetries. Pearson's correlations (r) were made to examine relationships between the anthropometric measures and both the average and peak instantaneous GRFv asymmetry. For these correlations, the anthropometric measures were normalized to the subject's standing height converted to a percentage and the overall averages of the GRFv asymmetries across set and repetition were used. Finally, intraclass correlations using Chronbach's Alpha were used to establish day-to-day repeatability with those that performed the procedures (anthropometric measurements and squat protocol) a second time. The GRFv asymmetry repeatability was assessed within the first set (both R1 and R2 and R7 and R8) and within the overall averages combining R1 and R2 and R7 and R8 of all sets. Statistical tests were conducted in PASW version 18.0 (SPSS, Inc., Chicago, IL, USA) with significance set at p ≤ 0.05. Because of the limited data available relative to the effect of fatigue on GRFv related functional asymmetries, an a priori power estimate to determine sample size was not performed. However, actual statistical power (Pr) for each variable was calculated and reported.


All subjects reported their right leg and arm as their preferred kicking leg and throwing arm, respectively, and completed all repetitions of all sets at the 90% of their 8RM load. There were only slight differences in anthropometrics between contralateral sides of the body, with thigh length and circumference greater on the right leg (Table 1). All anthropometrics yielded significant correlations when comparing left to ride sides (r ≥ 0.964, p < 0.001).

Table 1
Table 1:
Anthropometric measurements.*

The 2 × 5 MANOVA assessing differences in time to complete each repetition (R1 and R2, R7 and R8), and each set (1-5), had no significant repetition by set interaction (p = 0.229, Pr = 0.425) (Figure 3). However, there was a main effect for repetition with later repetitions taking longer to complete than initial repetitions within a single set (p < 0.001, Pr = 0.989). There was no main effect for set in time to complete repetitions (p = 0.782, Pr = 0.102).

Figure 3
Figure 3:
Whole group (n = 17) average time to complete repetitions 1 and 2 (R1 and R2) and 7 and 8 (R7 and R8) of each set with significant main effects occurring for repetition but not set. Error bars indicate first SD.

The absolute average GRFv asymmetry of the whole group (n = 17) was 3.6 ± 2.7% (range: 1.0–8.5%) for R1 and R2 of set 1 and 3.7 ± 2.6% (range: 0.3–8.4%) for R7 and R8 of set 1. Five subjects began R1 and R2 of set 1 with more weight on their right leg, whereas 12 began with more weight on their left leg. Eight of the 17 subjects switched sides bearing the greatest load during at least 1 of their 10 repetition data points across sets compared to R1 and R2 of their first set. Within these 8 subjects, 3 began R1 and R2 of set 1 with an absolute average GRFv asymmetry level <1.7%.

The absolute peak instantaneous GRFv asymmetry of the whole group (n = 17) was 3.0 ± 2.2% (range: 0.2–6.4%) for R1 and R2 of set 1 and 2.8 ± 2.7% (range: 0.2–7.7%) for R7 and R8 of set 1. Four subjects began R1 and R2 of set 1 with more weight on their right leg, whereas the other 13 began with more weight on their left leg. Nine of the 17 subjects switched sides bearing the greatest load during at least 1 of their 10 repetition data points across sets compared to R1 and R2 of their first set. Within these 9 subjects, 5 began R1 and R2 of set 1 with an absolute peak GRFv asymmetry level <1.7%.

As with time to complete repetitions, there were no repetition × set interactions in any of the analyses of GRFv asymmetries (p ≥ 0.214, Pr ranging from 0.105 to 0.279), nor were there any main effects for set (p ≥ 0.146, Pr ranging from 0.127 to 0.511). Whole-group analysis of absolute average GRFv asymmetry, where a reduction in value would indicate movement toward being more symmetric or possibly switching to the opposite leg they started with more weight on, failed to produce a significant main effect for repetition (n = 17, p = 0.600, Pr = 0.080) as did absolute peak instantaneous GRFv asymmetry (n = 17, p = 0.229, Pr = 0.172). When again assessed for absolute average GRFv asymmetry, with the initially highly symmetric subjects (±1.7%) removed, there was a main effect for repetition (n = 12, p = 0.044, Pr = 0.546), with subjects becoming more symmetric with fatigue at the end of a set, but this did not carry over to absolute peak instantaneous asymmetry (n = 12, p = 0.266, Pr = 0.188) (Figure 4).

Figure 4
Figure 4:
Whole group, with initially symmetric subjects removed: (A) absolute average vertical ground reaction force (GRFv) asymmetry (n = 12) and (B) absolute peak instantaneous GRFv asymmetry (n = 12). A significant main effect for repetition existed within the average absolute GRFv asymmetry. Error bars indicate first SD.

When examined to explore for a general shift toward the left or right foot in the whole group, with those that were initially highly symmetric removed, there was no main effect for repetition in the average %L–R GRFv asymmetry (n = 12, p = 0.788, Pr = 0.057). However, there was a significant main effect in the peak instantaneous %L–R GRFv asymmetry (n = 12, p = 0.042, Pr = 0.554), where again the group became more symmetric with fatigue (Figure 5). When assessing those individuals who began with more weight placed on their left side, with those that were initially highly symmetric removed, there was no main effect for repetition in average %L–R GRFv asymmetry (n = 9, p = 0.212, Pr = 0.224) or peak instantaneous %L–R GRFv asymmetry (n = 10, p = 0.080, Pr = 0.422) (Figure 6). When assessing those individuals who began with more weight placed on their right side, with those that were initially highly symmetric removed, a significant main effect existed for repetition in average %L–R GRFv asymmetry (n = 3, p = 0.036, Pr = 0.740) with movement toward symmetry but not for peak instantaneous %L–R GRFv asymmetry (n = 2, p = 0.540, Pr = 0.068) (Figure 7).

Figure 5
Figure 5:
Whole group, with initially symmetric subjects removed: (A) average %L–R vertical ground reaction force (GRFv) asymmetry (n = 12) and (B) peak instantaneous %L–R GRFv asymmetry (n = 12). A significant main effect for repetition existed within the peak instantaneous %L–R GRFv asymmetry. Error bars indicate first SD.
Figure 6
Figure 6:
Whole group, with initially symmetric subjects removed, of those individuals who began with more weight on their left side (A) average %L–R vertical ground reaction force (GRFv) asymmetry (n = 9) and (B) peak instantaneous %L–R GRFv asymmetry (n = 10). Error bars indicate first SD.
Figure 7
Figure 7:
Whole group, with initially symmetric subjects removed, of those individuals who began with more weight on their right side (A) average %L–R vertical ground reaction force (GRFv) asymmetry (n = 3) and (B) peak instantaneous %L–R GRFv asymmetry (n = 2). A significant main effect for repetition existed within the average %L–R GRFv asymmetry. Error bars indicate first SD. Note, the large SD of R7 and R8 of the peak instantaneous %L–R GRFv asymmetry graph was cropped (set 4 R7 and R8 ± 6.8%).

In the correlations of %L–R GRFv asymmetry with left minus right anthropometric differences normalized to the subject's standing height, only the thigh length and calf circumference yielded significant correlations to overall average %L–R GRFv asymmetry (Table 2). Although statistically significant, the 90% confidence intervals of the correlations were relatively large, ranging from −0.80 to −0.21 for the thigh length and −0.77 to −0.14 for the calf circumference. The only significant correlation for peak instantaneous %L–R GRFv asymmetry was with calf circumference (Table 2). Again, the 90% confidence interval for this correlation was relatively large, ranging from −0.77 to −0.14.

Table 2
Table 2:
Anthropometric correlations with GRFv asymmetry.*

All anthropometric measurements taken from the left and right sides of the body and height and weight were highly repeatable from day to day on the 8 subjects that returned for a third visit (Table 3). However, intraclass correlations of left minus right side anthropometric differences taken on different days only revealed significant correlations for foot length and calf circumference (Table 3). All time and GRFv asymmetry assessments were highly correlated from visit to visit (Table 4).

Table 3
Table 3:
Repeatability assessment of anthropometrics: Cronbach's alpha.*
Table 4
Table 4:
Repeatability assessment of squat protocol: Cronbach's alpha.*


The main goal of this investigation was to assess the effect of fatigue on functional asymmetries by examining differences in average and peak GRFv under each foot during the free-weight barbell back squat exercise. Secondary goals were to assess the relationship of the GRFv asymmetries with lower extremity anthropometric asymmetries and the day-to-day repeatability of all measures. Before discussing the findings related to these goals, it is important to address the presence of functional asymmetries within this group performing squats relative to others that have been examined.

As expected, even in this healthy population, functional GRFv asymmetries were typically present, which is similar to others using a crate lifting task (16), standing (23,24), static lifting (24), and sit-to-stand (15). Specific to the free-weight barbell back squat exercise, Newton et al. (19) reported significant differences in peak and average GRFv between strength based assignment of dominant and nondominant legs in collegiate women softball players. The only contradictory findings to the general consensus that GRFv asymmetries exist in healthy populations were the results by Song et al. (28) who discovered symmetric lower extremity joint kinetics in older adults while performing a body weight squat. The fact that the subjects only performed 3 trials and only the peak GRFv was analyzed could have played a role in not finding differences as could the type of analysis performed. If approximately half of the subjects favor one leg and the others favor the contralateral leg, the average could be apparent symmetry. Typically, as found in our study, people place more weight on the nonkicking leg, as it is known as the stance leg (7). Thus, it is reasonable to accept that individuals who identify their right leg as their preferred kicking leg would place more weight on their left leg (stance leg) and thus display greater GRFv on the left side when performing tasks.

When assessing the effects of fatigue of the squat protocol on GRFv asymmetries, there was a tendency for subjects to either stay the same or become slightly more symmetric both in the average GRFv and peak instantaneous GRFv. The changes with fatigue were extremely acute and only existed within a set, not across sets. The observed changes in average GRFv asymmetry appear to be heavily driven by those that started on their preferred kicking leg (right leg) at the beginning of the first set, though not entirely, because this was a small subset of the population. Conversely, it appears those that started on their nonkicking leg (left leg) drove the changes in peak instantaneous GRFv asymmetry. Why there was a difference between those responsible for the change in average asymmetry vs. peak instantaneous asymmetry is not clear. However, it is anticipated that the average level of asymmetry is more important than the peak instantaneous asymmetry, because the limb is exposed to the average force much longer than the instantaneous.

Previous research has examined the effects of fatigue on various jump tasks, such as landing from a jump (20,27), performing a drop jump (18), performing a maximal vertical jump (25), and single-leg hop testing (1), but is limited on the effects of fatigue during the squat exercise and how fatigue influences asymmetries. The majority of the literature examines fatigue in the context of reduced forces, joint moments and angles, jump height, or jump distance, and does not compare side-to-side differences to assess asymmetries. Therefore, it is difficult to relate the findings of this study to the work of others. It is, however, warranted to discuss the findings of Carpes et al. (3), as they revealed that pedaling asymmetries decreased as crank torque and cycling exercise intensity (% O2max) increased. Intensity relative to % O2max may then be a valid surrogate for fatigue, which is in line with the findings of the present study. This conclusion is based on the premise that as fatigue increases, maximum force capacity decreases. As a result, performance of a constant submaximal effort task eventually becomes a maximum effort task under extreme fatigue. Effectively, the null results reported by Flanagan and Salem (6) with the squat exercise at different levels of load, also supports our findings that asymmetries do not increase with fatigue.

It should, however, be noted that Oda and Moritani (20) demonstrated a difference in force decline, which was greater for the left arm, when comparing right and left arms of right-handed oarsmen who performed a sustained maximal voluntary elbow flexion exercise. Differences in external power output between the arms, which were greater for the left, of swimmers during swim bench exercises that consistently breathe to the same side have also been observed (22). Further, the magnitude of difference in bilateral arm power increased as subjects reached exhaustion. These groups (oarsmen and swimmers), however, were expected to be asymmetric based on their training and baseline measurements. This fact is reflected in the work by Krawczyk et al. (13) who measured 134 elite male athletes from 9 different sports. Their results suggest that asymmetries present in athletes are because of long-term, intense training and constant performance in sport-specific techniques and movements, which may then result in bone or muscle hypertrophy differences between limbs.

Despite the fact that the previously mentioned findings were on the upper extremities, it is reasonable to believe that such differences may exist in the lower extremities as well. Perhaps becoming more symmetric with fatigue in this study was a result of use of a bilateral exercise, which necessitated joint effort from both lower extremities to maintain balance simultaneously. Simon and Ferris (26), in a study assessing lower limb isometric extensions on an exercise machine, discovered that activating the legs at the same time while performing bilateral movements lead to asymmetric limb forces, concluding that asymmetries are neurologically based, rather than mechanically based. The greater differences seen at the start of a set in this study may have been because of subjects figuring out how to activate the lower limbs at the same time, while getting better at it through the course of a set, where greater symmetry was seen, thus a sort of “practice effect” occurred. This may also be why it did not carry over from set to set when a break was inserted but fatigue not fully recovered.

Because differences were seen within sets, but not across sets, perhaps if the rest time between sets was reduced, changes across sets and not just within repetitions may appear. Clearly the findings indicate that fatigue was induced within a set, as supported by the time to complete repetitions but did not heavily transfer from set to set. Adequate rest was given between sets as evidenced by the fact that time to complete repetitions did not increase across sets and all subjects completed all 8 repetitions of all 5 sets. The fact that subjects fatigued during a set, evidenced by taking more time to complete R7 and R8 compared to R1 and R2, and became more symmetric, relates to a finding by Koh et al. (12) who found a larger bilateral deficit when subjects were instructed to generate force rapidly rather than gradually. Perhaps the increased repetition time lead to decreased asymmetries in this study, as subjects generated force at a slightly reduced pace.

The use of a ±1.7% asymmetry cutoff for the present investigation was chosen arbitrarily to exclude those individuals who were highly symmetric to begin and not expected to change. It has no known relationship to performance or injury risk. Although methods of calculating and reporting asymmetries vary, levels ranging from 15 to 20% have been established and reported as clinically relevant bilateral strength asymmetries and increase risk for injury (9,10,32). Relationships of functional asymmetries to decrements in performance are limited. However, Yoshioka et al. (32), in a computer simulation of jumping, reported no jump height decrements with a 10% strength asymmetry but expected there would be with 20%. The use of a cutoff to exclude subjects with minimal differences between right and left sides has been performed elsewhere in the literature to determine if discrepancies between limbs exist (26).

Regarding anthropometric measurements, the study population was highly symmetric. Although statistical significance was found for thigh length and circumference, the magnitude of the differences was quite small and most likely not affecting the functional asymmetries observed. Repeatability of the anthropometric measurements was quite high from visit to visit, although not at the level to consistently reproduce side-to-side differences. This is most likely because the differences between the right and left sides were so small that any day-to-day variation in the measurement would affect the value. The relative lack of differences from side to side of the subjects suggests that a population was selected that had no anthropometric reason to expect significant levels of asymmetric function. This is supported by the low correlations of anthropometric differences to GRFv asymmetries. Of the variables that were significantly correlated, the confidence intervals were relatively large.

Within our investigation, asymmetries were relatively consistent for each subject during the workout and were highly repeatable for those that returned for the third visit. This is important to note because this protocol revealed that asymmetries, for the most part, are not random and maintain a consistent expression within and between visits. Maines and Reiser (16) also found consistent bilateral differences in GRFv when lifting a crate repeatedly from the floor within a visit and Reiser et al. (23) reported day-to-day weight-bearing repeatability during quiet stance. Because of the inconsistent nature of the changes in asymmetry level that did occur within and across sets of this study, they are most likely attributable to balance-related issues, because small corrections are required to prevent toppling forward, backwards, left, and right. With highly symmetric subjects, a small adjustment could cause a switch from more weight on one side to the other, which was observed in subjects in both the highly symmetric to start group (±1.7%) and the slightly more asymmetric group. This was evidenced by the fact that switching sides was an anomaly, rather than typical of subjects, indicating that if one starts with more weight placed on one foot at the beginning of the set, they were likely to keep weight on that same foot throughout the duration of the set. Furthermore, the high degree of consistency within a visit and across visits suggests that bar loading and placement is also highly consistent.

One of the limitations of this study was the relatively small sample size with an extremely homogenous population. This prohibited the ability to assess subjects that were left-handed and left-footed. Furthermore, statistical power was low in most cases. However, because side-to-side asymmetries were quite low in both anthropometric and functional variables, it would take many more subjects to improve the power level without any anticipated change in conclusion. More asymmetries might be found to statistically decrease with fatigue, but the finding that they do not increase and are consistent from day-to-day is most important. This is due to the fact that increases in asymmetry with fatigue would be a cause for concern both in terms of training stimulus and risk for injury. Also, if asymmetries were highly variable from day to day, there would be less concern for chronic exposure to asymmetric loads in those that were highly asymmetric.

Finally, additional data collection and analysis may have shed more light on the reason for functional asymmetries. Lower extremity kinematics and kinetics and muscle activation data would have allowed for further understanding of the sources of side-to-side differences. However, as Lawson et al. (14) and Stephens et al. (30) reported, finding consistent sources of asymmetry may be quite challenging in healthy groups that are not highly specialized in a task. Further, assessment of forces in more than one plane, such as medial, lateral, anterior, and posterior, may have allowed for additional understanding of asymmetries during the squat exercise. Horizontal force components are functionally important both in regard to performance and acute risk for injury, because balance and stability are an ongoing need during a free-weight barbell back squat. However because of their role in balance, they lend themselves to being highly variable from repetition to repetition and from person to person. This is supported by Maines and Reiser (16) and Giakas et al. (8) who reported high variability in these directions in tasks requiring balance. Furthermore, the horizontal forces are of a relatively low magnitude compared to the vertical forces. Therefore, based on variability and magnitude, the horizontal force components would not be expected to play a major role in unequal development.

Practical Applications

Low levels of relatively consistent functional asymmetries exist in healthy, recreationally trained men and women while performing a free-weight barbell back squat. Because the levels of asymmetry are quite low, strength and conditioning professionals should not be deterred from using this exercise for bilateral development of the lower extremities. Furthermore, the fact that subjects did not change or became slightly more symmetric with fatigue supports their value for bilateral development in healthy subjects with no known reason for being asymmetric. Although unilateral training may not be warranted based on these results for strength development, unilateral training may still be desirable for development of single-leg coordination and the development of muscles not specifically targeted by bilateral squats, such as those required for medial/lateral control of the hip, knee, and ankle. For those that may be predisposed to functional asymmetries because prior injury or development from asymmetric use, unilateral training is still highly advised. Finally, care should be taken when expanding these results beyond the free-weight barbell back squat exercise. Machine-based exercises, such as the Smith Machine, may not produce the same results because of the absence of a balance component. At the other end of the spectrum, when complete freedom of movement is allowed, such as during sport play, a person may favor one side of the body more readily, allowing asymmetries to increase with fatigue. However, the possibility of this has not been examined.


The authors thank all the volunteer subjects who freely gave their time and effort to this unfunded study. No professional relationships exist with any of the companies whose products were used. Results do not constitute endorsement of the products used by the authors or of the NSCA.


1. Augustsson, J, Thomee, R, Linden, C, Folkesson, M, Tranberg, R, and Karlsson, J. Single-leg hop testing following fatiguing exercise: Reliability and biomechanical analysis. Scand J Med Sci Sports 16: 111–120, 2006.
2. Blustein, SM and D'Amico, JC. Limb length discrepancy: Identification, clinical significance, and management. J Am Pediatr Med Assoc 75: 200–206, 1985.
3. Carpes, FP, Rossato, M, Faria, IE, and Bolli Mota, C. Bilateral pedaling asymmetry during a simulated 40-km cycling time-trial. J Sports Med Phys Fitness 47: 51–57, 2007.
4. Chandler, TJ and Stone, MH. The squat exercise in athletic conditioning: A review of the literature. Natl Strength Cond Assoc J 13: 51–60, 1991.
5. Evans, RC. Illustrated Essentials in Orthopedic Physical Assessment. St. Louis, MO: Mosby-Year Book Inc, 1994.
6. Flanagan, SP and Salem, GJ. Bilateral differences in the net joint torques during the squat exercise. J Strength Cond Res 21: 1220–1226, 2007.
7. Gabbard, C and Hart, S. A question of foot dominance. J Gen Psychol 123: 289–296, 1996.
8. Giakas, G, Baltzopoulos, V, Dangerfield, PH, Dorgan, JC, and Dalmira, S. Comparison of gait patterns between health and scoliotic patients using time and frequency domain analysis of ground reaction forces. Spine 21: 2235–2242, 1996.
9. Impellizzeri, FM, Rampinini, E, Maffiuletti, N, and Marcora, SM. A vertical jump force test for assessing bilateral strength asymmetry in athletes. Med Sci Sports Exerc 39: 2044–2050,2007.
10. Knapik, JJ, Bauman, CL, Jones, BH, Harris, JM, and Vaughan, L. Preseason strength and flexibility imbalances associated with athletic injuries in female collegiate athletes. Am J Sports Med 19: 76–81, 1991.
11. Kobayashi, Y, Kubo, J, Matsuo, A, Matsubayashi, T, Kobayashi, K, and Ishii, N. Bilateral asymmetry in joint torque during squat exercise performed by long jumpers. J Strength Cond Res 24: 2826–2830, 2010.
12. Koh, TJ, Grabiner, MD, and Clough, CA. Bilateral deficit is larger for step than for ramp isometric contractions. J Appl Physiol 74: 1200–1205, 1993.
13. Krawczyk, B, Sklad, M, Majle, B, and Jackiewicz, A. Lateral asymmetry in upper and lower limb measurements in selected groups of male athletes. Biol Sport 15: 33–38, 1998.
14. Lawson, BR, Stephens, TM, DeVoe, DE, and Reiser, RF. Lower-extremity bilateral differences during step-close and no-step countermovement jumps with concern for gender. J Strength Cond Res 20: 608–619, 2006.
15. Lundin, TM, Grabiner, MD, and Jahnigen, DW. On the assumption of bilateral lower extremity joint moment symmetry during the sit-to-stand task. J Biomech 28: 109–112, 1995.
16. Maines, JM and Reiser, RF. Ground reaction force bilateral asymmetries during submaximal sagittal plane lifting from the floor. Int J Ind Ergon 36: 109–117, 2006.
17. McLean, SG and Samorezov, JE. Fatigue-induced ACL injury risk stems from a degradation in central control. Med Sci Sports Exerc 41: 1661–1672, 2009.
18. Moran, KA, Clarke, M, Reilly, F, Wallace, ES, Brabazon, D, and Marshall, B. Does endurance fatigue increase the risk of injury when performing drop jumps? J Strength Cond Res 23: 1448–1455, 2009.
19. Newton, RU, Gerber, A, Nimphius, S, Shim, JK, Doan, BK, Robertson, M, Pearson, DR, Craig, BW, Häkkinen, K, and Kraemer, WJ. Determination of functional strength imbalance of the lower extremities. J Strength Cond Res 20: 971–977, 2006.
20. Oda, S and Moritani, T. Cross-correlation of bilateral differences in fatigue during sustained maximal voluntary contraction. Eur J Appl Physiol 70: 305–310, 1995.
21. Pappas, E, Sheikhzadeh, A, Hagins, M, and Nordin, M. The effect of gender and fatigue on the biomechanics of bilateral landings from a jump: Peak values. J Sports Sci Med 6: 77–84, 2007.
22. Potts, AD, Charlton, JE, and Smith, HM. Bilateral arm power imbalance in swim bench exercise to exhaustion. J Sports Sci 20: 975–979, 2002.
23. Reiser, R, Dalton, E, and Pault, J. Trial number and duration effects on standing weight-bearing asymmetry measures. Biomed Sci Instrum 44: 262–267, 2008.
24. Rocheford, EC, DeVoe, DE, and Reiser, RF. Effect of previous unilateral injuries on ground reaction force bilateral asymmetries during static lifting and standing. J Hum Mov Stud 51: 403–424, 2006.
25. Rodacki, ALF, Fowler, NE, and Bennett, SJ. Vertical jump coordination: Fatigue effects. Med Sci Sports Exerc 34: 105–116, 2002.
26. Simon, AM and Ferris, DP. Lower limb force production and bilateral force asymmetries are based on sense of effort. Exp Brain Res 187: 129–138, 2008.
27. Smith, MP, Sizer, PS, and James, CR. Effects of fatigue on frontal plane knee motion, muscle activity, and ground reaction forces in men and women during landing. J Sports Sci Med 8: 419–427, 2009.
28. Song, J-E, Flanagan, SP, Admon, C, Wang, M-Y, and Salem, GJ. Lower-extremity joint kinetics in older adults are symmetric while performing the squat exercise. Med Sci Sports Exerc 35: S347, 2003.
29. Sparto, RJ, Parnianpour, MP, Reinsel, TE, and Simon, S. The effect of fatigue on mulitjoint kinematics, coordination, and postural stability during a repetitive lifting test. J Orthopaed Sports Phys Ther 25: 3–12, 1997.
30. Stephens, TM, Lawson, BR, DeVoe, DE, and Reiser, RF. Gender and bilateral differences in single-leg countermovement jump performance with comparison to a double-leg jump. J Appl Biomech 23: 190–202, 2007.
31. Tomkinson, GT, Popovic, N, and Martin, M. Bilateral symmetry and the competitive standard attained in elite and sub-elite sport. J Sport Sci 21: 201–211, 2003.
32. Yoshioka, S, Nagano, A, Hay, D, and Fukashiro, S. The effect of bilateral asymmetry of muscle strength on jumping height of the countermovment jump: A computer simulation study. J Sports Sci 28: 209–218, 2010.

symmetry; functional asymmetry; resistance training

© 2011 National Strength and Conditioning Association