1. Human flexor reflex (HFR) responses were elicited during ergometer cycling in neurologically intact humans with the objective of understanding the influence of lower limb muscle activity on phase-dependent reflex modulation during movement. The experimental setup permitted control over background muscle activity and stimulus intensity without significantly interfering with the cycling motion. 2. All experiments involved cycling on an ergometer at a set rate and workload. A 333-Hz, 15-ms pulse train of electrical stimulation was randomly delivered to the skin over the tibial nerve at the ankle at selected lower limb positions. In the first group of experiments, subjects were stimulated at six cycling phases while pedaling with normal, phasic ankle activity (free-form cycling). The second and third group of experiments involved stimulation under static limb positioning conditions and during active pedaling while subjects were asked to maintain a consistent background level of isolated tibialis anterior (TA) or soleus (SOL) electromyographic (EMG) activity. 3. Control criteria were established to assure similar isolated muscle EMG levels and sensory stimulation intensities throughout the experiments. With the aid of the application of a lower extremity brace and visual EMG feedback, SOL and TA activity were confined by the subject to a narrow range during the task of cycling. Stimulus consistency was achieved through maintenance of flexor hallucis brevis M-waves to within an envelope encompassing the mean value +/- 5% of the maximum M-wave amplitude in all experimental conditions. 4. When the subject's limb was statically positioned, the HFR responses in the SOL muscle showed no significant changes in pattern when compared at various limb positions. During cycling with consistent SOL activity, a response waveform pattern of early-latency-long-duration depression was followed by a later-latency facilitation response in all positions except the initial power phase. The initial power phase was characterized by an additional early-latency facilitation in all but one subject. 5. In the TA muscle response, no change in onset latency (57.5 +/- 0.8 ms, mean +/- SD), waveform pattern, or response amplitude (7.9 +/- 1.1% maximal voluntary contraction, mean +/- SD) was observed during static limb positioning. Significant increases in response amplitude (P < 0.05) coupled with significant increases (9.2 ms, P < 0.05) in onset latency were seen during the transition from the recovery phase to the power phase during cycling. In addition, there was no correlation between the prestimulation baseline level and the onset latency during controlled TA cycling activity conditions.(ABSTRACT TRUNCATED AT 400 WORDS)