Human runners exhibit a least variable gait speed.

This study characterises the relationship between gait variability and speed in runners using data from trunk accelerations in each axis. Twelve participants of varying fitness ran on the treadmill with three sessions of six randomly ordered self-selected speeds. A VO2max test was conducted on the fourth session. Running gait was tracked with inertial sensors. The occurrence of a mid-range speed was analysed for the anterior-posterior, vertical and lateral directional coefficient of variation (CV) of root mean square (RMS) acceleration data. One participant with noisy gait signals was omitted. The results show all remaining participants consistently showed significant quadratic U-shaped relationships between vertical RMS CV acceleration and speed. Neither anterior-posterior nor lateral RMS CV acceleration were clearly related to speed. These least variable gait speeds were similar to estimates of optimal speed derived from minimum cost of transport with speed. In conclusion, there exists a mid-range speed for each runner with the least variable gait in the vertical direction, and this occurred significantly more often than would be expected by chance (P < 0.05). However, there are no prominent patterns for the anterior-posterior and lateral directions. This finding supports anecdotal evidence from runners and coaches concerning gait consistency.