High-resistance interval training produces substantial gains in sprint and endurance performance of cyclists in the competitive phase of a season. Here, we report the effect of changing the cadence of the intervals. We randomized 18 road cyclists to 2 groups for 4 weeks of training. Both groups replaced part of their usual training with 8 30-minute sessions consisting of sets of explosive single-leg jumps alternating with sets of high-intensity cycling sprints performed at either low cadence (60-70 min(-1)) or high cadence (110-120 min(-1)) on a training ergometer. Testosterone concentration was assayed in saliva samples collected before and after each session. Cycle ergometry before and after the intervention provided measures of performance (mean power in a 60-s time trial, incremental peak power, 4-mM lactate power) and physiologic indices of endurance performance (maximum oxygen uptake, exercise economy, fractional utilization of maximum oxygen uptake). Testosterone concentration in each session increased by 97% +/- 39% (mean +/- between-subject SD) in the low-cadence group but by only 62% +/- 23% in the high-cadence group. Performance in the low-cadence group improved more than in the high-cadence group, with mean differences of 2.5% (90% confidence limits, +/-4.8%) for 60-second mean power, 3.6% (+/-3.7%) for peak power, and 7.0% (+/-5.9%) for 4-mM lactate power. Maximum oxygen uptake showed a corresponding mean difference of 3.2% (+/-4.2%), but differences for other physiologic indices were unclear. Correlations between changes in performance and physiology were also unclear. Low-cadence interval training is probably more effective than high-cadence training in improving performance of well-trained competitive cyclists. The effects on performance may be related to training-associated effects on testosterone and to effects on maximum oxygen uptake.