During dynamic exercise, a large fall in systemic vascular resistance occurs. Arterial pressure (AP) is, however, maintained through a combination of central command and neural activity from muscle afferents that adjust the autonomic outflow to the circulation. How these signals are calibrated to provide accurate regulation of AP remains unclear. This study tests the hypothesis that the calibration can be 'learnt' through feedback from the arterial baroreceptors arising over multiple trials of exercise. Eight healthy subjects undertook three different protocols in random order. The test protocol consisted of 7 days' training, when subjects were exposed on 70 occasions to 4 min of exercise (50% of maximal oxygen uptake capacity) paired with neck suction (-40 mmHg) to mimic an excessive rise in AP at the carotid baroreceptors with exercise. Two control protocols involved training with either exercise or neck suction alone. No significant changes in mean AP, diastolic AP or heart rate during normal exercise were detected following training with any protocol. However, the rise in systolic AP with exercise was attenuated by an average of 7.3 +/- 2.0 mmHg (mean +/- s.e.m., P < 0.01) on the first and second days following training with the test protocol, but not with either control protocol (P < 0.05 for difference between protocols, ANOVA). In conclusion, this study failed to show that mean AP during normal exercise could be reduced through prior conditioning by overstimulation of the baroreceptors during exercise. However, a reduction in systolic AP was observed that suggests the presence of some plasticity within the autonomic response, consistent with our hypothesis.