METHODS:
Cyclists performed a 25-kJ sprint test, an incremental test to determine peak aerobic power (PP) and a simulated 40-km time-trial on a Kingcycle ergometer. They were then randomly assigned to one of five types of interval-training session: 12x30 s at 175% PP, 12x60 s at
100% PP, 12x2 min at 90% PP, 8x4 min at 85% PP, or 4x8 min at 80% PP. Cyclists completed 6 sessions over 3 wk, in addition to their usual aerobic base training. All laboratory tests were then repeated.
RESULTS:
Performances in the time trial were highly reliable when controlled for training effects (coefficient of variation = 1.1%). The percent improvement in the time trial was modeled as a polynomial function of the rank order of the intensity of the training intervals, a procedure validated by simulation. The cubic trend was strong and statistically significant (overall correlation = 0.70, P = 0.005) and predicted greatest enhancement for the intervals performed at 85% PP (2.8%, 95% CI = 4.3-1.3%) and at 175% PP (2.4%, 95% CI = 4.0-0.7%). Intervals performed at
100% PP and 80% PP did not produce statistically significant enhancements of performance. Quadratic and linear trends were weak or insubstantial.