Journal Logo


Analysis of Pacing Strategy Selection in Elite 400-m Freestyle Swimming


Author Information
Medicine & Science in Sports & Exercise: November 2012 - Volume 44 - Issue 11 - p 2205-2212
doi: 10.1249/MSS.0b013e3182604b84
  • Free


In sporting events where the sole aim of the athlete is to cover a set distance in the fastest possible time, it is generally accepted (2,3,17,18) that the adoption of a suitable pacing strategy is an important determinant of success. A pacing strategy can be defined as the manipulation of power output (PO) over an exercise bout, so as to balance energy expenditure and speed in a way that will allow completion of the activity to the best of the individuals’ capacity (17). Therefore, pacing is inextricably linked to the individuals’ “intelligent” regulation of the rate of fatigue development over the exercise.

Abbiss and Laursen (2) have previously described and defined six different pacing strategies common in sport performance, and different pacing profiles may be more or less beneficial depending on the sport, event, or distance. However, under stable conditions, it has been suggested that a constant PO (even pacing) is optimal for prolonged (>2 min) events such as running, cycling, rowing, and skiing (2). Indeed, more successful athletes have been shown to use a more even pacing strategy (22,32). The theory of the benefits of even pacing is primarily based on critical power models (12) and mathematical laws of motion (9) that state that even minor fluctuations in speed can result in greater energy cost. Despite this, the incidence of both parabolic and positive pacing strategies is also high in prolonged exercise (2).

Although the effect of different pacing strategies on performance has received considerable attention in many sports, including cycling (3,6), running (21,31), speed skating (14), and rowing (13), little attention has been paid to pacing strategy selection and swimming. Optimal pacing in swimming is made highly relevant due to the increased resistive forces brought about by the water. As swimming velocity increases, frontal water resistance increases disproportionately (4); therefore, any fluctuations in velocity could create higher relative energy costs (10,25). Because of the higher resistive forces in swimming (than that in other sports) that change disproportionately as a result of changes in velocity, pacing may be all the more important because less frequent changes in velocity may reduce the energy cost of overcoming drag. Indeed, “variable” pacing has previously been shown to be ineffective in swimming (27), and instead, a more even pacing strategy (3) with an initial fast start (5,7) may be more appropriate. De Koning et al. (8) support this concept, stating that PO lost to the environment (that may be minimized in swimming by reducing changes in velocity) is a key factor in determining the pacing pattern in a given race. Despite the apparent importance of pacing selection to performance in swimming, further complexities such as the effect of competitors and race tactics may cause athletes to deviate from what may be more biomechanically, mathematically, and physiologically optimal.

There are higher resistive forces (e.g., drag) in swimming than that in many other sports (10), which may allow for a greater significance of pacing on performance outcome. Therefore, anything that helps reduce these forces may also effect pacing strategy selection. Indeed, swimming performance is partly determined by the body drag of the swimmer, which is composed of friction, pressure, and wave-making resistance (20). Because friction and pressure resistance are highly related to the flow conditions and body characteristics, anything that may serve to reduce these will also reduce drag and the energy lost to the environment for a given velocity. Different swimsuit designs and materials, such as whole torso suits and polyurethane high tech suits (PhtS), have been shown to reduce drag, increase buoyancy, and improve performance by as much as 10% (19,24). Therefore, it may also be expected that the reduced metabolic cost for a given exercise intensity that such suits provide may also influence pacing by making any changes in velocity less costly in terms of energy lost to overcoming increased drag.

Therefore, the aim of the current study was to investigate the prominence of different pacing strategies in elite swimming competitions and their relationship with performance outcome. To achieve this, 264 swims from national and international elite competitions were analyzed for pacing strategy, completion time, sex, and type of swimming suit worn. It was hypothesized that there would be no interaction between pacing strategy selection and race performance.



After ethical approval from the Institutional Ethics Committee (University of Bedfordshire), race analysis reports for two hundred sixty-four 400-m freestyle swims (male = 147, female = 117) of swimmers competing in the finals at British and Australian Championships, International Invitational Meets, European Championships, World Championships, and Commonwealth and Olympic Games in the period 2003–2010 were obtained from British Swimming. All races were swum in long-course (50-m) pools. All British Swimming race analysis reports included the completion time, the split times, and the ratio of completion time to the current world record (at the time of the event). The information retrospectively obtained from the race analysis reports was subsequently collated and analyzed by the investigators. Information on the swimmers’ swimming suit design (PhtS or non–high tech suit (NhtS)) was retrospectively added by the investigators by corresponding the race analysis report date and the swimmer to the competition.


British Swimming race analysis reports provided mean velocity (m·s−1) for each swimmer over 16 sections (pacing sectors) of the 400-m race. Pacing sectors were identified by dividing lengths into two segments, excluding the first 15 m at the start and the last 5 m and the first 10 m in each length, because of the turn. The final 5 m of the race was included. The breakdown of the 16 pacing sectors is depicted in Figure 1. Average swimming velocity for each of these sectors was calculated on the basis of change in position (distance) and elapsed time (time), measured from the head of the swimmer through video analysis. For the video analysis, a video camera was placed at the 25-m mark on the poolside and used to film the swimmer of interest. All video recordings were completed by British Swimming and mean velocities logged by a skilled scientist. All the raw data obtained from the British Swimming race analysis reports were subsequently collated and analyzed by the current authors. Because of the fixed position of the camera, some level of angle distortion would be expected, and this may have created a small level of error in the speed, distance, and time equations used to calculate velocity for each pacing sector. This level of distortion would be greatest furthest from the camera (i.e., the far lane and at the ends of the pool) but minimal where the angle of the pacing sector and camera were perpendicular. Where the distortion was greatest, it is estimated that the largest angle of distortion would have been <24°, creating a distance error of <0.485 m and a possible timing error of <0.29 s (0.12% of finishing time, based on a mean race velocity of 1.65 m·s−1). However, because differences in finishing time were between 1 and 3 s, and pacing sectors were allocated by differences of >1% of finishing time, 0.12% (<0.29 s) represents a small level of error, and differences found between strategies are still likely meaningful.

Pacing sectors in a long-course (50-m) pool used to establish changes in velocity for the pacing analysis. Pacing sector 1 is between 15 and 25 m of race distance, pacing sector 2 is between 25 and 40 m of race distance, and so on. Swimmers’ distance completed was recorded from the head.

Data analysis.

The classification of pacing strategy for each race profile was determined through an algorithm, built by the current investigators, using OpenOffice 3.2.1 Calc (Oracle Corp., Redwood Shores, Redwood City, CA). Although the constructed algorithm requires validation, its use still represents an objective means of classifying pacing strategies to preset criteria determined by investigators and based on previous research (2). Before running the race profile through the algorithm, an average race velocity was calculated so that the velocity in each pacing sector could be expressed in relation to overall race velocity (normalized mean velocity). For example, if an average race velocity was 1.65 m·s−1, a velocity of 1.62 m·s−1 for a particular pacing sector would be 98% of average race velocity. This approach to expressing pacing strategy as the difference between current velocity and overall mean velocity is well accepted (2). The algorithm allocated a level of conformity for the normalized mean velocity in each pacing sector of each swim, against the predominant pacing strategies previously identified in the literature (2). These key pacing strategies were modeled into 16 normalized mean velocity sectors so that they corresponded with the pacing data obtained from the race analysis reports used in the current study. Therefore, each pacing sector in each modeled strategy had a normalized mean velocity range that the observed swims could “fit” into. These modeled strategies and range of normalized mean velocity for each pacing sector are displayed in Figure 2. For example, a normalized mean velocity of 98% for pacing sector 3 would allocate one conformity point for an even pacing strategy. The algorithm worked through all 16 pacing sectors, allocating a conformity point each time a pacing sector in a swim “matched” a modeled pacing sector. The pacing strategy that was allocated the most conformity points (up to a maximum of 16) resulted in the swim profile being classified with that pacing strategy (e.g., even pacing strategy). In some instances, a pacing sector from a swim may match the same sector from more than one modeled strategy. However, as a conformity point for each sector was obtained, each complete swim received more or less conformity points for a particular modeled strategy. For example, although the normalized mean velocity of a swim may match the initial pacing sectors for the modeled parabolic, fast-start-even and positive pacing strategies (and therefore, each modeled strategy would acquire a conformity point for these sectors), one strategy would ultimately acquire more matched sectors (and thus conformity points) overall. The algorithm allowed the investigators to input the swimmer’s velocity for each sector, and the algorithm would calculate normalized velocity and allocate conformity points for each pacing sector. In the case of two pacing strategies being allocated the same total number of conformity points (e.g., 10 conformity points each), the normalized velocity range for each pacing sector of these schemas was narrowed in steps of 0.2% and the algorithm applied again until one schema had more conformity points. Variable pacing strategies were identified visually, because these types of profile could take many forms that could not all be modeled and classified by an algorithm.

From the six pacing schemas identified by Abbiss and Laursen (2) (parabolic (A), fast-start-even (B), positive (C), negative (D), even (E), and variable (F)) only three were frequently used by the elite swimmers (A–C). Each strategy was modeled by expressing velocity as a percentage of race mean velocity (normalized mean velocity) (12). Normalized mean velocity was modeled according to each pacing strategy defined by Abbiss and Laursen (2) for each of the 16 pacing sectors, with a range above and below the mean, expressed in the figure by error bars. Pacing sectors from each observed swim were compared against the modeled pacing profile sectors depicted in this figure via an algorithm. The modeled strategy that had the highest conformity (i.e., the most sectors that fit within the error bars) to the observed swim would then be labeled as such. Variable pacing strategies were identified visually (F) irrespective of sex; 89 swims were classified as parabolic, 120 as fast-start-even, 31 as positive, 4 as negative, 0 as even, and 20 as variable.

Statistical analysis.

Descriptive data are presented as mean ± SD. All analyses were conducted using SPSS version 16.0 (Chicago, IL), and significance was accepted when P < 0.05. Data were analyzed using a three-way ANOVA (pacing strategy × sex × swimming suit) in an unrelated design.


Distribution of pacing strategies.

Table 1 shows the incidence of different pacing strategy selection for swims analyzed in this study. Fast-start-even and parabolic pacing profiles were used the most, with parabolic profiles preferred by a greater extent by males. Irrespective of sex or the swimming suit worn, positive, negative, and variable pacing strategies were infrequently used. On no occasion was a variable pacing strategy used. Of the 264 swims analyzed, no swimmers selected an even pacing strategy, and the 24 negative (n = 4) and variable (n = 20) profiles were excluded from the subsequent statistical analysis because of insufficient sample size.

Number of swims conforming to each pacing profile from all selected competitions.

Main effect of pacing strategy, sex, and swimming suit.

Swimmers selecting a fast-start-even pacing strategy produced performances that were an average of 96.08% ± 2.12% (95% confidence interval (CI) = 95.86%–96.81%) of the world record time (%WR) (equating to a mean time of approximately 3:48.4 ± 4.66 s), with swimmers selecting parabolic pacing strategies producing similar performances (96.04% ± 2.2%) (95% CI = 95.87%–96.79%) (equating to a mean time of approximately 3:48.7 ± 4.84 s). Positive strategies (95.4% ± 2.19%) (95% CI = 95.12%–97.52%) were generally further from the world record (equating to a mean time of approximately 3:50.12 ± 4.82 s) than both fast-start-even and parabolic pacing strategies (Fig. 3). Although this difference was insignificant (F2,228 = 1.00, P > 0.05), the absolute time difference of this percentage difference between fast-start-even and positive pacing equated to approximately 1.7 s. Although negative and variable strategies were removed from the statistical analysis because of the small number of samples, negative pacing produced performances that were generally a high %WR (98.02% ± 0.57%), as did variable pacing (96.6% ± 1.75%).

Performance outcome for pacing strategy, sex, and swimsuit type. Performance outcome is expressed as performance time as a percentage of the world record time at the time of competition. *Denotes significant difference (P < 0.05).

There was a significant main effect for sex (F1,228 = 11.1, P < 0.01), with females achieving significantly closer performances (96.59% ± 2.07%) (95% CI = 96.33%–97.87%) to the female world record (equating to a mean time of approximately 4:08.12 ± 4.97 s) than male swimmers to the male world record (95.45% ± 2.1%) (95% CI = 95.10%–96.10%) (equating to a mean time of approximately 3:50.01 ± 4.62 s).

A significant main effect for the type of swimsuit (PhtS and NhtS) was also observed (F1,228 = 6.59, P < 0.05). Swimmers using a PhtS achieved significantly and consistently closer performances (96.74% ± 2.16%) (95% CI = 96.10%–97.76%) to the world record (equating to a mean time of approximately 3:47.12 ± 4.75 s) than those who used a NhtS (95.73% ± 2.01%) (95% CI = 95.39%–96.09%) (equating to a mean time of approximately 3:49.39 ± 4.42 s). Mean performances, dependent on relation to pacing strategy, sex, and swimsuit, are shown in Table 2.

Performance outcome for each pacing strategy displayed as percentage of the current world record time.

Interaction effects for pacing strategy × sex × swimming suit.

Although no significant interaction effect was observed between pacing strategy and sex (F2,228 = 2.72, P > 0.05), there appeared to be a greater sex difference in %WR when swimmers paced positive than when they paced fast-start-even or parabolic (Fig. 4).

Interaction effect for pacing strategy and sex for performance outcome. Although not significant, there was a trend toward greater sex differences when pacing positive.

No interaction effect for pacing strategy and swimsuit (F2,228 = 0.18, P > 0.05) or sex and swimsuit (F1,228 = 0.01, P > 0.05) were observed.


This study was designed to establish the prominence of different pacing strategies in elite 400-m freestyle swimming and their relationship with performance outcome by retrospectively analyzing swims from recent elite national and international meets. The primary finding of the study was that although fast-start-even and parabolic pacing profiles were clearly favored by competitors, no single pacing strategy appeared to exert a significant influence on performance time. Despite this, pacing strategies in elite swimming that use a fast start appear to be selected with significantly more frequency than slow or even starts. Because all the swimmers analyzed in this study were elite athletes, this would suggest that elite athletes may naturally select the most effective pacing strategy. To the authors’ knowledge, this is the first large-scale, field-based study on elite freestyle swimmers where velocity has been obtained in high frequency (every 6% of total race distance). Therefore, this study acknowledges the recommendations by Foster et al. (10), who state that split times should be measured every 5%–10% of the race when investigating pacing and performance.

Although the effect of pacing strategy selection on performance outcome was found to be nonsignificant, the mean difference between competitors pacing positive and fast-start-even or parabolic was approximately 0.7% WR, equating in absolute terms to approximately 1.7 s. Where the difference between medal winners at international-level 400-m freestyle races is frequently <1 s, the functional performance difference observed in the current study could be meaningful. Furthermore, the observation that positive pacing resulted in worse performances supports previous literature (2,22,32), and thus, the lack of statistical significance of this effect may solely be a factor of a small number of samples (31 positive vs. 120 fast-start-even) in this pacing group. It may also be that because of the inefficient nature of this pacing strategy, elite athletes make a tactical decision not to select it, which would explain the low incidence of this pacing profile in elite competition. However, on examination of the range of performance times across the three most prominent strategies, it becomes clear that, despite different means, ranges overlap. Therefore, it may be that selection of one of these three strategies does not pose a significant hindrance in elite 400-m swimming performance. The finding that the majority of swimmers adopted a fast-start-even or parabolic pacing strategy (as opposed to variable or even strategies), and that these performances were consistently closer to the world record (although not significant), seems to suggest that these strategies are a factor in optimal performance in 400-m freestyle swimming and that they are sub/consciously “chosen” by experienced competitors. However, it could also be the case that more talented swimmers naturally select these strategies, which would also explain the better performances when using these profiles. Given that parabolic and fast-start-even strategies are very similar, apart from the final section, it may be that their form of work-rate distribution yields positive physiological or biomechanical advantages. In support of this, Bishop et al. (5) have previously shown that fast-start-even pacing produces better performances than even pacing and attributed this to initial breakdown of PCr and an increase in O2 without a further accumulation of O2 deficit. This finding is supported by Jones et al. (15), who found that an initial higher PO led to increased oxidative contribution to energy turnover, thereby improving exercise tolerance and sparing some anaerobic work capacity for later in the exercise. Sandals et al. (23) have further shown that positive pacing leads to a significantly greater %VO2max during a race. Therefore, it may be that in events of similar short-medium duration, regardless of sport, particular forms of pacing produce the physiological benefits alluded to. Indeed, de Koning et al. (7) have shown that an initial high PO followed by a constant PO is the optimal strategy for 4000-m cycling performance. Because a 4000-m cycling time trial lasts approximately 4:20 min (i.e., similar in duration to the 400-m freestyle swim), this provides further support for the findings of this study. However, it should be noted that very high initial POs are required in cycling to overcome inertia. Because this requirement is largely escaped in swimming (because of the dive), the higher initial work rate in swimming would also be reflected by a proportional higher speed (whereas in cycling, a high PO would be observed, but a lower speed than the mean race velocity). In contrast, however, Foster et al. (11) stated that an even pacing strategy is optimal for cycling performances lasting 2–3 min, because even minimal changes in velocity during the starting sections can have negative consequences on overall race performance. Given the increased resistive forces in swimming, this recommendation appears relevant; however, Foster et al. (11) did not observe any changes in O2 uptake kinetics, B[La] accumulation, or O2 deficit, which may be attributed to relatively lower fast start PO differences. Therefore, it may be that a short-duration, high-PO fast start to overcome initial resistive forces, followed by minimal changes in work rate, may provide both biomechanical and physiological benefits for events of this duration. The main difference between the fast-start-even and parabolic pacing strategy is the presence/absence of the sudden increase in work rate in the final section of the race. This phenomenon has been termed the end spurt or end sprint and has previously been suggested as evidence for a protective mechanism of central control (21,31). The presence of this end sprint in 89 of the 264 swims suggests that these swimmers maintained some level of physiological reserve that they were able to use when the race end point was more proximal. This may represent a reserve capacity that can only be accessed when premature fatigue is unlikely and peripheral fatigue signaling can be overridden, a change in race tactics as a result of a competitor’s pacing strategy or an error in the previous sections of the swim, which caused a lower than optimal work rate (and a subsequent ability to increase it). A curious observation in the relative success of the different strategies was that variable and negative pacing (despite only being selected by a very small number of swimmers) actually produced the fastest swim times (96.6% ± 1.75% WR and 98.02% ± 0.57% WR, respectively). Both of these strategies would increase the relative time spent accelerating and, with the increased resistive forces experienced in swimming, would usually be considered energetically inefficient (4,9,10,25). There may be several explanations for this particular observation: 1) the fastest swimmers selected the most inefficient strategy and therefore offset the pacing disadvantage. 2) Variable and negative pacing elicited some kind of tactical advantage. 3) Pacing strategy selection offers little energetic, physiological, or biomechanical advantage in swimming races of this distance. Ultimately, it is likely that an elite swimmer selects a pacing strategy that they are most comfortable with for that particular race, which may vary from day to day and is likely dependent on race tactics and the other competitors (27).

It is curious that relatively few competitors used an even pacing strategy, given the attention and recognition this type of pacing has received in previous literature (2,22,32). However, differences in laboratory-based (3,7) and field-based (13) study, modality of exercise (9,13,14,31), exercise duration (18,31), effect of competition (13), fixed (15) or self-paced (17) exercise, conceptual paradigm (physiology, biomechanics, and psychology) (6,7,15), and course topography/environmental conditions (3,6) are likely behind observed differences in literature regarding pacing. These differences are difficult to reconcile, and because it will often depend on the research question and the sport, it is important to recognize that different pacing patterns may be more or less beneficial depending on each of the factors given (8). However, one important analysis feature that is often overlooked, but can have significant importance in understanding pacing, is the frequency of data acquisition (10). Studies (13,31) often use low-frequency work-rate data points that are too insensitive to identify spontaneous changes in pacing behavior, which have been suggested to be vitally important (16,30) to performance. This lack of sensitivity also results in a reduced number of potential pacing profiles that can be defined and thus cause confusion between what strategy is actually used. For example, in contrast to the current study, Tucker et al. (31) found that 26 different world records in 800-m track running were achieved by adopting a positive pacing strategy. However, the work rates were only taken every 50% of the race, except in 12 races where split times were taken every 25%. This low resolution would result in any fast start being masked and potentially cause confusion with fast-start-even pacing, which may explain the differences between the studies. Indeed, using a higher frequency when analyzing 5000-m performance, Tucker et al. (31) found that the greatest number of world records was achieved when athletes ran a parabolic pacing profile. This highlights the need for researchers to capture work-rate data points in high frequencies, if possible, between every 5%–10% of the total distance, as suggested by Foster et al. (10) and as performed in the current study.

Although the significant effect for sex may be initially surprising, given the standardization of the data to current sex-specific world records at the time of the event and its contradiction to previous data that were normalized in a similar manner (13), it is highly likely that the observed sex difference is due to a greater number of females competing at similar levels to the world record. Whereas between 2003 and 2010, the female world record changed six times, the male world record only changed once, which indicates the apparent sex difference was likely influenced by an individual higher world record performance of a single male, compared with a larger and more competitive group of females. Although a standardization of data is always required to investigate sex differences, because of differences in physiological characteristics among male and female swimmers effecting performance outcome, there is a need for a more reliable normalization method, which is not effected by individual performances. Females appeared to select a fast-start-even pacing strategy more frequently than males (53% vs. 39% of swims, respectively), although the performance difference between sex for this strategy was minimal (1.04% WR) and nonsignificant. However, the sex performance difference between positive pacing (despite similar selection frequency, females = 10% and males = 13% of swims, respectively) was more pronounced (2.59% WR), and this difference approached significance (P = 0.68). Therefore, it may be that elite female swimmers more successfully used positive pacing, and although this strategy was not selected most frequently by females, it did seem to produce (although not significantly so) the fastest mean performances in this group (Fig. 4). Given that it should be expected that elite athletes would naturally select the “fastest” pacing strategy, it would be more logical that positive pacing in females would be selected more frequently. Speculatively, it may be that this kind of pacing is physiologically and tactically more optimal (5,15,23,26), but the greater level of discomfort this may in turn produce (as a result of greater metabolite accumulation) may make it a greater risk and thus result in fewer swimmers selecting it (or completing it effectively).

The finding in the current study that the use of a PhtS resulted in a significant improvement in performance outcome is in agreement with previous studies. According to Tomikawa and Nomura (28), the PhtS enhances buoyancy and reduces water resistance, which subsequently improves performance time. As with the previously discussed sex difference, it is likely that the improved performances when using the PhtS (despite the normalization method) is due to comparative differences between the use of the PhtS and NhtS against the same world record time. Comparisons between competitions when the PhtS were banned and Olympic Games and World Championships where performers would usually be “peaking” and produce better performances were done. Because PhtS increase propulsion efficiency (29) and decreases water resistance (28), it may have been expected that changes in velocity during a swim may be less detrimental to overall efficiency, and therefore, the negative effects of variable pacing may be less apparent. However, no such pacing strategy and suit interaction were observed, and therefore, it appears as though the use of a PhtS improves performance without changing pacing.

It is important to note that this study has provided a retrospective analysis of pacing profiles in elite swimming. No intervention was performed, and no physiological or psychological data were collected; therefore, mechanisms underpinning the observed effects remain speculative. On the basis of previous laboratory-based research, it is likely that improved O2 kinetics (5,15), sparing of anaerobic work capacity (15), and reductions in oxygen deficit (5), in combination with various biomechanical factors (4), are responsible for certain pacing strategies being more frequently selected by elite performers than others. However, these explanations do not take into account how an athlete may intelligently regulate their pace over the course of a race. Indeed, a reduced oxygen deficit is a likely consequence of a fast start (5) rather than its cause. The low incidence of positive pacing, and prevalence of an end sprint, is suggestive of a system of central control, which allows access to a physiological reserve once race end point is close enough to minimize risk of premature fatigue or physical harm as a result of increasing work rate (17,18,21,31). Indeed, the observation that a range of pacing strategies were used in these elite competitions, despite the fixed conditions and high experience levels, suggests that pacing is more than a case of simply swimming as fast as possible. Whether this is dependent on race tactics, central regulation, or individual preference for strategy remains to be elucidated. Despite the lack of mechanistic explanation of this article, this is one of a small number of studies that provides an insight into pacing behavior in elite competition and, to the authors’ knowledge, the only such large-scale study in elite freestyle swimming.

It should be noted that the fixed camera position used in the current study limits the level of accuracy for the calculation of velocity for each pacing sector. Although the level of error is likely <0.12% and that this is less than the observed difference in pacing and performance, it is recommended that future research uses a camera setup where multiple cameras, or a moving camera, allows a perpendicular view of the swimmer crossing each pacing sector.

In conclusion, this study demonstrates that fast-start-even and parabolic pacing strategies are used more frequently in elite 400-m swimming freestyle competitions. Because of their higher incidence of use by elite performers, these particular strategies may yield better performances than other strategies, such as negative, even, or variable pacing, which are commonly reported in other sports. The athlete preference for fast-start-even and parabolic pacing occurs regardless of sex or swimming suit design, and functional difference between these strategies during competition appears to be minimal. Athletes and coaches should consider adopting pacing training sessions into their programs, particularly for developing athletes, to accommodate the growing literature support for the efficacy of certain pacing strategies in time-based events.

The authors would like to thank British Swimming for granting access and permission to use their race report data.

No funding was received for the conduct of this study.

The results of the present study do not constitute endorsement by the American College of Sports Medicine.


1. Abbiss CR, Laursen PB. Models to explain fatigue during prolonged endurance cycling. Sports Med. 2005; 35 (10): 865–98.
2. Abbiss CR, Laursen PB. Describing and understanding pacing strategies during athletic competition. Sports Med. 2008; 38 (3): 239–52.
3. Atkinson G, Brunskill A. Pacing strategies during a cycling time trial with simulated headwinds and tailwinds. Ergonomics. 2000; 43 (10): 1449–60.
4. Batchelor G. An Introduction to Fluid Dynamics. Cambridge: Cambridge University Press; 1967. p. 246.
5. Bishop D, Bonetti D, Dawson B. The influence of pacing strategy on VO2 and supramaximal kayak performance. Med Sci Sports Exerc. 2002; 34 (6): 1041–7.
6. Cangley P, Passfield L, Carter H, Bailey M. The effect of variable gradients on pacing in cycling time-trials. Int J Sports Med. 2011; 32 (2): 132–6.
7. de Koning JJ, Bobbert MF, Foster C. Determination of optimal pacing strategy in track cycling with an energy flow model. J Sci Med Sport. 1999; 2 (3): 266–77.
8. de Koning JJ, Foster C, Lucia A, Bobbert MF, Hettinga FJ, Porcari JP. Using modeling to understand how athletes in different disciplines solve the same question: swimming vs. running vs. speed skating. Int J Sports Physiol Perform. 2011; 6: 276–80.
9. di Prampero PE, Cortili G, Mognoni P, Saibene F. Equation of motion of a cyclist. J Appl Physiol. 1979; 47 (1): 201–6.
10. Foster C, Schrager M., Snyder AC, Thompson NN. Pacing strategy and athletic performance. Sports Med. 1994; 17 (2): 77–85.
11. Foster C, Snyder AC, Thompson NN, Green MA, Foley M, Schrager M. Effect of pacing strategy on cycle time trial performance. Med Sci Sports Exerc. 1993; 25 (3): 383–8.
12. Fukuba Y, Whipp BJ. A metabolic limit on the ability to make up for lost time in endurance events. J Appl Physiol. 1999; 87 (2): 853–61.
13. Garland SW. An analysis of the pacing strategy adopted by elite competitors in 2000 m rowing. Br J Sports Med. 2005; 39 (1): 39–42.
14. Hettinga FJ, De Koning JJ, Schmidt LJ, Wind NA, Macintosh BR, Foster C. Optimal pacing strategy: from theoretical modelling to reality in 1500-m speed skating. Br J Sports Med. 2011; 45 (1): 30–5.
15. Jones AM, Wilkerson DP, Vanhatalo A, Burnley M. Influence of pacing strategy on O2 uptake and exercise tolerance. Scand J Med Sci Sports. 2008; 18 (5): 615–26.
16. Lander PJ, Butterly RJ, Edwards AM. Self-paced exercise is less physically challenging than enforced constant pace exercise of the same intensity: influence of complex central metabolic control. Br J Sports Med. 2009; 43: 789–95.
17. Mauger AR, Jones AM, Williams CA. Influence of feedback and prior experience on pacing during a 4-km cycle time trial. Med Sci Sports Exerc. 2009; 41 (2): 451–8.
18. Mauger AR, Jones AM, Williams CA. Influence of exercise variation on the retention of a pacing strategy. Eur J Appl Physiol. 2010; 108 (5): 1015–23.
19. Mollendorf JC, Termin AC, Oppenheim E, Pendergast DR. Effect of swim suit design on passive drag. Med Sci Sports Exerc. 2004; 36 (6): 1029–35.
20. Naemi R, Easson WJ, Sanders RH. Hydrodynamic glide efficiency in swimming. J Sci Med Sport. 2010; 13: 444–51.
21. Noakes TD, Lambert MI, Hauman R. Which lap is the slowest? An analysis of 32 world mile record performances. Br J Sports Med. 2009; 43 (10): 760–4.
22. Padilla S, Mujika I, Angulo F, Goiriena JJ. Scientific approach to the 1-h cycling world record: a case study. J Appl Physiol. 2000; 89 (4): 1522–7.
23. Sandals LE, Wood DM, Draper SB, James DV. Influence of pacing strategy on oxygen uptake during treadmill middle-distance running. Int J Sports Med. 2006; 27 (1): 37–42.
24. Starling RD, Costill DL, Trappe TA, Jozsi AC, Trappe SW, Goodpaster BH. Effect of swimming suit design on the energy demands of swimming. Med Sci Sports Exerc. 1995; 27 (7): 1086–9.
25. Swain DP. A model for optimizing cycling performance by varying power on hills and in wind. Med Sci Sports Exerc. 1997; 29 (8): 1104–8.
26. Thompson KG, MacLaren DP, Lees A, Atkinson G. The effect of even, positive and negative pacing on metabolic, kinematic and temporal variables during breaststroke swimming. Eur J Appl Physiol. 2003; 88: 438–43.
27. Thompson KG, MacLaren DP, Lees A, Atkinson G. The effects of changing pace on metabolism and stroke characteristics during high-speed breaststroke swimming. J Sports Sci. 2004; 22 (2): 149–57.
28. Tomikawa M, Nomura T. Relationships between swim performance, maximal oxygen uptake and peak power output when wearing a wetsuit. J Sci Med Sport. 2009; 12 (2): 317–22.
29. Tomikawa M, Shimoyama Y, Nomura T. Factors related to the advantageous effects of wearing a wetsuit during swimming at different submaximal velocity in triathletes. J Sci Med Sport. 2008; 11 (4): 417–23.
30. Tucker R, Bester A, Lambert EV, Noakes TD, Vaughan CL, St Clair Gibson A. Non-random fluctuations in power output during self-paced exercise. Br J Sports Med. 2006; 40 (11): 912–7.
31. Tucker R, Lambert MI, Noakes TD. An analysis of pacing strategies during men’s world-record performances in track athletics. Int J Sports Physiol Perform. 2006; 1 (3): 233–45.
32. Wilberg RB, Pratt J. A survey of the race profiles of cyclists in the pursuit and kilo track events. Can J Sport Sci. 1988; 13 (4): 208–13.


©2012The American College of Sports Medicine