The purpose of this study was to examine the effect of ankle position on the human ankle stretch reflexes during tonically-maintained contractions over most of the range of motion. The ankle was placed at randomly selected mean positions. Target levels of triceps surae (TS) or tibialis anterior (TA) tonic contractions were generated while the ankle was displaced by small amplitude, stochastic perturbations. System identification techniques were used to identify the stretch reflex dynamics at each combination of tonic level and ankle angle. As shown previously, the TS stretch reflex was characterized by an unidirectional, velocity-sensitive impulse response function whereas the TA stretch reflex was characterized by a linear impulse response function between ankle velocity and TA EMG. TS stretch reflexes showed a strong dependence on ankle position while TA stretch reflexes did not. Thus the TS stretch reflex magnitude increased greatly as the ankle was progressively dorsiflexed. In contrast, ankle mean position had only a minor effect on the TA stretch reflex magnitude. Our results indicate that the position-dependent facilitation of the TS stretch reflex is not due to changes in the level of skeletal motoneuron excitability. Rather, this effect may be accounted for by mechanisms that modulate the efficacy of the stochastic ankle perturbation. Such mechanisms could include position-induced: modulation of monosynaptic and polysynaptic afferent inputs to skeletal motoneurons, alterations in the extent of fusimotor drive and changes in the transmission of the joint perturbation to spindle receptors. Such mechanisms are discussed in terms of the differences between TS and TA stretch reflexes. Finally, the functional significance of position-dependent reflex responses are considered.