We have examined the force-velocity characteristics of tetanically activated human adductor pollicis working in vivo, in the fresh and fatigued states. The increase in force in response to stretch was divided into two major components. The first, steady, component persisted after the stretch and is concluded not to be a function of active cycling cross-bridges because it was not affected by either the velocity of the stretch or the level of muscle activation. The origin of the second, transient, component of the increased force seen during stretch is consistent with cross-bridge activity since it increased with increasing velocity of stretch and was proportional to the level of activation. It is likely that both components of the stretch response make a significant contribution to muscle performance when acting to resist a force. For the fastest stretch used, the contributions of cross-bridge and non-cross-bridge mechanisms were equal. For the slowest stretch, lasting 10 s and over the same distance, the force response was attributed almost entirely to non-cross-bridge mechanisms. As a result of acute fatigue (50 % isometric force loss) there were only small reductions in the non-cross-bridge component of the force response to stretch, while the cross-bridge component decreased in absolute terms. The transient component of the stretch response increased as a result of fatigue, relative to the isometric force, while the force during shortening decreased. The results are consistent with a decrease in cross-bridge turnover in fatigued muscle.