1. The relationship between work-rate and the antihypertensive effect of exercise in hypertensives, and the mechanism of that effect, were investigated by a crossover clinical trial. 2. Ten mild hypertensives were randomly divided into two groups. One group performed low work-rate exercise (LWE) on a cycle ergometer for 10 weeks (blood lactate threshold; approximately 50% of maximum oxygen consumption [Vo2max]). After a 10 week interval without exercise training, these subjects were then switched to a high work-rate exercise (HWE) regimen (4 mmol/L of blood lactate; approximately 75% of Vo2max) for another 10 weeks. In the other group, the order of exercise training was reversed. Since two patients withdrew from the protocol during HWE periods, statistical analysis was performed on the data from the remaining eight patients. There were no order effects observed in any of the data from the two groups. 3. During both LWE and HWE, resting blood pressure (BP) fell significantly after the initiation of exercise therapy (P < 0.05). Furthermore, the overall effects of 10 weeks of LWE and HWE on BP were not significantly different. 4. The work-rate at the lactate threshold, which reflects physical fitness, had increased significantly by 16 W (P < 0.01) after the LWE period and by 11 W (P < 0.01) after the HWE. 5. During the LWE period, changes in haemodynamic and humoral variables were not significant, except for a reduction in plasma norepinephrine at week 10 (P < 0.05). In the HWE period, changes in haemodynamic and humoral variables were not significant.(ABSTRACT TRUNCATED AT 250 WORDS)