Objective. To compare the maximum gait speed of stroke subjects attained during treadmill and overground in stroke subjects and to identify the temporal-distance determinants of the maximal gait speed. Methods. Ten individuals with hemiparetic gait deficits and whose walking speeds ranged between 0.24 m/s and 0.82 m/s participated. Five healthy age-matched controls were also tested to provide comparative data for the gait speed transfer between the 2 modes of locomotion. Following a brief habituation process to walking on the treadmill, subjects were tested while walking at comfortable and maximal speeds on the treadmill and overground, in a random order. Main Outcome Measure. Self-selected comfortable and maximum gait speed and temporal-distance factors were acquired using a 6-camera Vicon™ motion analysis system and compared between treadmill and overground walking at a similar speed. Results. Overground walking resulted in higher maximal speeds (P < 0.001), greater stride lengths (P < 0.001), and a lower cadence (P < 0.02), as compared to tread-mill. The comfortable gait speed and the maximum stride length proved to be strong determinants for the maximal speed on both modes of locomotion (P < 0.01), but the maximum cadence was correlated to maximum speed only for overground locomotion (P < 0.05). Conclusions. Stroke subjects walked slower on the treadmill as compared to overground. They also used a different strategy to increase gait speed, relying mostly on increasing the stride length during treadmill ambulation.