Share this article on:

A kinematic analysis of high-speed treadmill sprinting over a range of velocities


Medicine & Science in Sports & Exercise: April 2002 - Volume 34 - Issue 4 - p 662-666

KIVI, D. M. R., B. K. V. MARAJ, and P. GERVAIS. A kinematic analysis of high-speed treadmill sprinting over a range of velocities. Med. Sci. Sports Exerc., Vol. 34, No. 4, pp. 662–666, 2002.

Introduction The purpose of this study was to measure changes in stride characteristics and lower-extremity kinematics of the hip and knee as a function of increasing treadmill velocity, at velocities ranging from submaximal to near maximal.

Methods: Six power/speed athletes experienced at sprinting on a treadmill performed trials at 70%, 80%, 90%, and 95% of their previous individual maximum velocity, with video data collected in the sagittal view at 60 Hz.

Results: Significant differences were seen in stride frequency (70%, 80%, P < 0.01; 90%, P < 0.05), stance time (70%, 80%, P < 0.01; 90%, P < 0.05) flight time (70%, P < 0.01; 80%, P < 0.05), hip flexion angle (70%, P < 0.01), hip flexion angular velocity (70%, P < 0.01), hip extension angular velocity (70%, 80%, P < 0.01), knee flexion angular velocity (70%, 80%, P < 0.01), and knee extension angular velocity (70%, P < 0.01), as compared with the near maximum (95%) velocity. Coefficient of variation (CV) values showed that the positional variables at the hip and knee were more variable at faster test conditions, indicating that kinematic changes occur as a function of increased treadmill velocity.

Conclusions The results indicated that at slower velocities, there were differences in the stride characteristics and lower-extremity kinematics while sprinting on a treadmill. As the velocity approached near maximum mechanical breakdown was seen, suggesting that velocities greater than 90% should be used selectively during treadmill training.

Sports Biomechanics Laboratory and Perceptual Motor Behavior Laboratory, University of Alberta, Edmonton, AB, CANADA

Submitted for publication January 2001.

Accepted for publication August 2001.

High-speed treadmills have become popular training tools for speed development. Capable of reaching speeds of 12.5 m·s−1, inclines up to 40°, and declines of 10°, they allow for closely controlled training under a wide range of conditions. These treadmills also place an emphasis on the movement velocity and movement specificity, which must be considered during training for speed development (4–6). Previous studies have shown the importance of specificity of training in sprinting, including overground studies (12), resistance-training methods (2,13), and in sprint-specific strength exercises (11).

Despite the high degree of specificity expected between overground and treadmill sprinting, it has been reported that there are kinematic differences between these two training methods. Frishberg (3) compared the sprint kinematics in five collegiate-level sprinters overground (mean velocity = 8.54 ± 0.09 m·s−1) and on a treadmill (mean velocity = 8.46 ± 0.13 m·s−1). No significant differences were seen in stride frequency, step length, support time, or flight time between the two conditions. Differences were observed in the lower extremity during the support phase. Specifically, the support thigh was more erect at contact and moved with a slower angular velocity, whereas the support shank was less erect at contact and moved with a greater range of motion and angular velocity while sprinting on a treadmill.

In Frishberg’s study, the participants completed a 91.44-m maximal run overground, with average horizontal velocity calculations at each 9.14-m interval. A treadmill sprint was then performed, with the investigators attempting to match the interval velocities as closely as possible. Data for both tests collected at 41.12–50.26 m. However, questions arise when using a treadmill in this manner. It is extremely difficult to create a smooth and consistent acceleration pattern as seen during a sprint race manually, and unnatural for the sprinter to adjust their cadence to a constantly changing treadmill velocity. The differences seen between overground and treadmill sprinting may not entirely be a result of the comparison between training methods but due to the difficulty of the treadmill test itself. An investigation into the sprinting on a constantly moving treadmill over a range of velocities would reveal any changes or mechanical breakdown associated with sprinting on a treadmill. Accordingly, it was the purpose of this study to measure changes in stride characteristics and lower-extremity kinematics as a function of increasing treadmill velocity, at velocities ranging from submaximal to near maximal.

Back to Top | Article Outline


Participants and experimental protocol.

Six participants (3 male, 3 female) were recruited for this study, all of whom competed nationally or internationally in power/speed events in athletics (200 m, 400 m, 400-m hurdles, and pole vault). These subjects were selected based on their familiarity with sprinting on the high-speed treadmill, having used it regularly for more than a year as part of their training program. All subjects volunteered for this study and gave informed written consent to serve as subjects. The Institutional Review Board approved procedures for the experiment and participant involvement. Mean height was 1.76 ± 0.01 m for the male subjects and 1.67 ± 0.05 m for the female subjects. The male and female subjects had a mean masses of 73.8 ± 5.6 kg and 59.1 ± 5.5 kg, respectively.

The testing took place during the competitive portion of the indoor season. After completing their individual warm-up, each subject was fitted with 2-cm reflective markers that were located on the joint centers of the hip (greater trochanter), knee (lateral epicondyle), ankle (lateral malleolus), heel (middle of calcaneus), and foot (base of fifth metatarsal) on the side facing the camera.

Trials were completed on an Acceleration® high-speed running treadmill (Frappier Acceleration Sports Training, Fargo, ND) set at 0° of inclination. Each participant completed runs of 70%, 80%, 90%, and 95% of their individual maximum velocity on the treadmill, which had previously been established during training. Trials were completed from the slowest to fastest velocity, as this is the manner in which the athletes were accustomed to using the treadmill. A trial consisted of the athlete preparing on the side of the treadmill while holding onto the support handles, and in one motion, jumping on and beginning to sprint. When the athletes established their strides, they removed their hands and maintained a stationary position on the treadmill in which there was no visible horizontal deviation of the torso for a period of 3–5 s, after which they were free to jump off. Subjects were also instructed to jump off the treadmill at any time if they did not feel comfortable or felt they were unable to continue. Video was also collected at 100% but was not included because the participants were unable to achieve and maintain a stationary position on the treadmill for a period of time that was sufficient for analysis. The treadmill was preset to a specific velocity for each individual before the trial; there was no acceleration or deceleration phase. A stride was defined as the time from ground contact of one foot to the next ground contact of the same foot. For each trial, three successive strides were analyzed. The participants were given sufficient rest between trials, approximately 6–8 min.

Back to Top | Article Outline

Data analyses.

Video data of the sagittal plane motion were collected using a Canon 8-mm video camera, which recorded at 60 Hz and a shutter speed of 1/2000. The camera was located such that the optical axis was perpendicular to the plane of movement. Data processing was completed using an Ariel Performance Analysis System. For each trial, the time-dependent coordinates of each landmark were smoothed using a low-pass digital filter with a cutoff frequency of 8 Hz to reduce small random errors that may have occurred during digitizing, without introducing systematic bias. The cutoff frequency was determined by inspection of the raw and filtered data and comparison between the respective power spectra. Absolute hip joint angles with respect to the vertical and relative knee angles were calculated (see Fig. 1). Finite differences were used to calculate linear and angular velocity data for each segment with hip extension, hip extension angular velocity, and knee flexion angular velocity defined as negative. Stance and flight times were measured in video frames and converted to time, with stride frequency calculated as the reciprocal of the sum of the stance and flight times. For all the dependent measures used, coefficients of variation (CV) were calculated.



Back to Top | Article Outline

Statistical analyses.

Descriptive statistics (means and SD) were calculated for variables selected for analysis based on previous kinematic treadmill and sprint studies (3,7,8,9,10). Differences between the test conditions were analyzed using a one-way repeated measures analysis of variance (ANOVA). Post hoc tests were used to determine where differences were seen among the 70%, 80%, and 90% velocities in comparison to the 95% test condition. Probability values less than 0.05 were taken to indicate statistical significance.

Back to Top | Article Outline


Mean treadmill velocities with standard deviations are reported in Table 1. Significant differences were seen between the velocities at 70–80% and 80–90% (P < 0.01), with no significant differences between the fastest two test conditions.

Table 1

Table 1

Back to Top | Article Outline

Stride characteristics.

Stride characteristics while treadmill sprinting over a range of velocities are seen in Table 2, with the means and SD for the kinematic variables analyzed at the selected test conditions. Mean stride frequency was significantly less at the 70% and 80% velocity than compared to the 95% velocity, with the stride frequency at 90% showing no significant differences. Mean stance time at 70%, 80% (P < 0.01), and 90% (P < 0.05) were significantly longer than at 95%. Mean flight time was longer at the 70% (P < 0.01) and 80% (P < 0.05) than compared to 95%, with no differences seen at 90%.

Table 2

Table 2

Stride frequency was seen to increase systematically as the treadmill velocity increased and was a result of decreases in both stance and flight phases. Luhtanen and Komi (7) showed increases in stride frequency as overground sprinting velocity increased from submaximal to maximal for male track and field athletes. These authors also reported decreases in flight and support time with increased velocity. These results are in agreement with those of the present study, showing that when there are increases in sprinting speed on either the treadmill or overground, there is an increase in stride frequency that is a result of a decrease in both flight and stance times.

Back to Top | Article Outline

Hip kinematics.

Table 3 shows the kinematics of the hip while sprinting on a treadmill over a range of velocities. There was a significantly smaller mean maximum angle of hip flexion angle at 70% velocity (P < 0.01), with 80% and 90% showing similar flexion angles to the near maximum. For mean maximum hip extension, there were no differences seen at any of the velocities analyzed. Hip flexion angular velocity was significantly slower at the 70% velocity (P < 0.01), with 80% and 90% showing no differences. Mean maximum hip extension angular velocity was significantly slower at 70% (P < 0.01) and 80% (P < 0.05), with no differences at 90%.

Table 3

Table 3

Peak hip flexion angle increased nonsignificantly as the treadmill velocity increased, reaching a maximum value at 90%. The 95% value, however, was slightly smaller than at 90%, which is not ideal. Hip flexion angular velocity showed a similar nonsignificant trend in which the highest angular velocity value was seen at the 90% velocity. Although the differences seen in these two variables were only statistically significant between slowest and fastest test conditions, the results are important from a practical and technical perspective. Mann and Herman (9) stated that maximizing the flexion angle of the hip is crucial in producing upper-leg angular velocity before and during ground contact. At near maximum treadmill velocities, the angle of hip flexion is limited because the sprinter is unable to continue increasing the hip flexion angular velocity of the recovery leg. At this high stride frequency, the sprinter cannot effectively recover the leg forward and must restrict the hip range of motion to maintain the high cadence and make proper ground contact.

Hip extension angular velocity was significantly smaller at the two slowest test conditions but was seen to increase nonsignificantly as treadmill velocity increased. The results of this study indicate that one of the strengths of training at near maximum velocities on a treadmill may be in increasing hip extension angular velocity. Frishberg (3), however, suggested that it is the moving treadmill that brings the support leg back under the body during the support phase of running, thereby reducing the energy requirement of the runner.

After toe-off, the maximum angle of hip extension showed no significant differences among the velocities tested, which is a desirable characteristic of sprinters. Mann (8) indicated that “good” sprinters minimize the upper leg rotation at full extension, which reduces ground contact time and makes leg recovery more efficient. Despite the increased hip extension angular velocity seen as the treadmill velocity increased, the sprinters were able to control the extension of the hip after toe-off, maintaining a good position for fast leg recovery.

Back to Top | Article Outline

Knee kinematics.

Kinematics of the knee are presented in Table 4. Neither maximum knee flexion nor knee extension at toe-off was significantly different from the 95% velocity for any of the test conditions. Mean maximum knee-flexion angular velocity differed significantly (P < 0.01) at the 70% and 80% velocities. Mean maximum knee-extension angular velocity was significantly slower at the 70% velocity (P < 0.01), with the 80% and 90% velocities showing no differences.

Table 4

Table 4

According to Mann (8), better sprinters minimize the angle of knee flexion during recovery to make the task faster and easier, as was seen in this study. At toe-off, no significant differences were seen in the angle of knee extension. This shows that as the treadmill velocity increases, sprinters attempt to limit the lower-leg range of motion to minimize the ground contact time and to enable the leg to recover fast enough for the next ground contact, a finding consistent with previous work (8).

Knee extension angular velocity is important in allowing the lower leg enough time to be able to produce sufficient knee flexion angular velocity at touchdown, which will reduce the forward braking force during the initial portion of ground contact (8). The angular velocity for knee extension during the slowest treadmill condition was significantly different from the maximum velocity analyzed. These differences match very closely with the peak hip flexion angle and the peak hip flexion angular velocity variables, suggesting that the ability of a sprinter to reduce braking forces during ground contact may be related to the ability to recover the leg forward. If the leg is recovered faster, the athlete will be in a better position to initiate the backward acceleration of the leg to ground contact.

Back to Top | Article Outline

Coefficient of variation.

Coefficient of variation (CV) has been used previously in gait studies as an indicator of the variation among individuals (1,14). CV was used in this study as an indicator of mechanical breakdown. Expressed as a percentage, the larger the CV the greater the variation seen among the subjects at a given treadmill velocity, thereby indicating a larger disparity in technique. Smaller CV values would be indicative of similarities across subjects and thus less technical breakdown.

CV values are reported in Tables 1–4 for each variable under each test condition. CV was similar for each treadmill velocity analyzed, which indicates that variation was consistent as velocities increased. For the stride characteristics, CV values increased for stance time, with consistent values for stride frequency and flight time for each test condition. From these results, we suggest that sprinters are better at adjusting their stride frequency and flight time to faster treadmill velocities but show more variability in their ability to adapt their support time.

At the hip and knee, similar patterns were seen in the CV values, in which the positional variables all showed increasing variability as the treadmill velocity increased. The most notable CV increase was seen in the angle of hip extension, which showed a relatively large deviation across conditions as the treadmill velocity increased, despite the fact that there were no significant differences among the means. These results indicate that as treadmill velocity increases, sprinters become less consistent in the angular positions they can achieve. Moreover, this may be an indication that significant mechanical changes occur as a function of increased treadmill velocity. The angular velocity values, however, all showed the greatest amount of variation at the 80% test condition. As the velocities reached near maximum, the variability decreased. This may be a result of changes as the sprinter goes from what can be described as “fast running” to “sprinting,” after which a more stable pattern of movement emerges.

The results of this study indicate that there are differences in most kinematic variables while sprinting on a high-speed treadmill at a range of velocities. Eight of 11 variables showed significant differences and were seen primarily at slower velocities. These findings have a number of direct implications for sprint training. Treadmill sprinting should not be performed at speeds of 80% or less, as the stride characteristics and hip and knee kinematics are considerably different from those of near maximum velocities. The angular velocity values at the hip and knee also show the greatest amount of variation at 80% after which the variation decreases, which may be an indicator that the sprinters are in a possible transition between sprinting and fast running. In addition, we feel that sprint mechanics on a treadmill may begin to breakdown as velocities reach near maximum. This was seen in the CV values for all of the positional variables at the hip and knee, which increased with faster treadmill velocities.

The authors believe that there are a number of benefits for training on a high-speed treadmill, including decreasing support and nonsupport time, and increasing hip extension angular velocity. However, given the differences seen at slower velocities and the mechanical breakdown found at near maximum velocities, we suggest that sprinters incorporating a treadmill as part of their training regimen use 90% of their individual maximum on the treadmill as their velocity for speed development. Speeds greater than 90% should be used selectively as they may develop unwanted technical adaptations and should be coupled with overground training to properly develop the physiological factors important to maximum speed sprinting. Further research is required into the mechanics of sprinting on a high-speed treadmill, in order to more fully understand the implementation of this training tool into the sprinter’s regimen.

The authors would like to thank Athletics Canada for its assistance in completing this project.

Address for correspondence: Mr. Derek Kivi, Sports Biomechanics Laboratory, Faculty of Physical Education, University of Alberta, Edmonton, AB, Canada, T6G 2H9; E-mail:

Back to Top | Article Outline


1. Challis, J. H. The variability in running gait caused by force plate targeting. J. Appl. Biomech. 17: 77–83, 2001.
2. Delecluse, C., H. Van Coppenolle, E. Willems, M. Van Leemputte, R. Diels, and M. Gordis. Influence of high resistance and high velocity training on sprint performance. Med. Sci. Sports. Exerc. 27: 1203–1209, 1995.
3. Frishberg, B. A. An analysis of overground and treadmill sprinting. Med. Sci. Sports Exerc. 15: 478–485, 1983.
4. Gauffin, H., J. Ekstrand, L. Arnesson, and H. Tropp. Vertical jump performance in soccer players. J. Hum. Mov. Stud. 16: 215–224, 1989.
5. Hakkinen, K., and P. V. Komi. Changes in electrical and mechanical behavior of leg extensor muscles during heavy resistance strength training. Scand. J. Sports. Sci. 7:55–64.
6. Kaneko, M., T. Fuchimoto, H. Toji, and K. Suei. Training effect of different loads on the force-velocity relationship and mechanical power output in human muscle. Scand. J. Sports. Sci. 5: 50–55, 1983.
7. Luhtanen, P., and P. V. Komi. Mechanical factors influencing running speed. In: Biomechanics VI-B, E. Asmussen and K. Jorgensen (Eds.). Baltimore: University Park Press, 1978, pp. 23–29.
8. Mann, R. V. Biomechanical analyses of the elite sprinter and hurdler. In: The Elite Athlete, N. K. Butts, T. T. Gushiken, and B. Zarins (Eds.). Jamaica, NY: Spectrum Publications, 1985, pp. 43–80.
9. Mann, R. V., and J. Herman. Kinematic analysis of Olympic sprint performance: men’s 200 metres. Int. J. Sports Biomech. 5: 151–162, 1985.
10. Mero, A., P. V. Komi, and R. J. Gregor. Biomechanics of sprint running: a review. Sports Med. 13: 376–392, 1992.
11. Mero, A, and P. V. Komi. EMG, force, and power analysis of sprint-specific strength exercises. J. Appl. Biomech. 10: 1–13, 1994.
12. Perez, H. R. Specificity of training: selective aspects of running. J. Sports Med. 21:68–73.
13. Wilson, G. J., R. U. Newton, A. J. Murphy, and B. J. Humphries. The optimal training load for the development of dynamic athletic performance. Med. Sci. Sports. Exerc. 25 (11): 1279–1285, 1993.
14. Winter, D. A. The Biomechanics and Motor Control of Human Gait: Normal, Elderly and Pathological, 2nd Ed. Waterloo, ON: University of Waterloo Press, 1991, p. 9.


© 2002 American College of Sports Medicine