1. Nine subjects received 6 baseline and 24 training sessions, each consisting of 250 elbow joint stretch perturbations into extension. The training sessions were designed to reduce the subjects' biceps brachii spinal stretch reflexes (SSR). Changes in longer-latency responses and short-latency brachioradialis responses were also monitored. Background electromyogram activity was recorded from the lateral head of triceps brachii during the biceps sampling intervals. These data were compared with those form 12 control subjects who received equal numbers of stretches over an extended baseline (i.e., without application of a training paradigm). 2. Training subjects reduced their mean biceps SSR responses by 24%, whereas control subjects increased their responses by 12% When changes in activity were grouped by sets of six consecutive sessions following the baseline interval, group-by-time interaction was observed. Training subjects showed significant reductions from baseline after the first raining set. The increased magnitude of biceps SSR for control subjects was significantly larger than baseline in sets 2-5. After the first training interval, all subsequent differences between groups were statistically significant. 3. The brachioradialis showed greater response variability, but these responses paralleled those seen in biceps brachii. The training group reduced their brachioradialis responses by 18%, whereas the control group increased their responses by 12%. Background activity recorded over the lateral head of triceps brachii during the biceps brachii SSR window became smaller in both groups. 4. Comparison of data between control subjects and subjects who had undergone biceps SSR up-training in previous studies suggests that the small increase seen among the present control subjects was probably not the result of a direct training effect. 5. Baseline differences in longer-latency response onset time were seen between groups. Over extended baseline sessions, longer-latency responses showed only a 1% onset, representing a change of 0.3 ms, in the control group. A 1.3% later onset, equaling a change of 1.1 ms, was observed over conditioning sessions in the training group. When analyzed by sets following baseline, neither group showed significant within- or between-group changes over time. 6. The magnitude of the longer-latency biceps brachii response showed 7% and 37% reductions for the control and training groups, respectively. No difference in set averages was seen within groups, but a significant difference was measured over sets 2-4 between groups. Reductions in the magnitude of longer-latency responses in training subjects were more obvious in later training sessions. 7. These data suggest that 1) the biceps brachii SSR can be reduced in human subjects; 2) concurrent changes are often observed in the synergist brachioradialis; and 3) the reduced biceps SSR magnitude is not linked to a compensatory increase in antagonist muscle activity. Reductions in the magnitude of longer-latency biceps brachii activity seen in parallel with the decreased SSR may imply that some degree of supraspinal processing is required to achieve this task.