The reduction in heart rate (HR) during the first minute of recovery immediately after a graded maximal exercise stress test (GXT) has recently been found to be a powerful and independent predictor of cardiovascular and all-cause mortality. Reduced vagal activity has been postulated as the cause, but this has not been proven in a population with slow HR recovery (HRR). To investigate autonomic contributions to HRR using time-frequency analysis in a group of individuals demonstrating slow HRR. HRR was defined as the difference in HR between peak exercise and 1 min later; a value < or = 18 bpm was set as threshold and considered abnormal. A modified continuous wavelet transform (CWT) was used to perform time-dependent spectral analysis during the baseline steady state and the following non-steady-state conditions created by GXT. This method provides dynamic measures of low-frequency (LF) and high-frequency (HF) peaks associated with autonomic activity. Individuals (N = 20) with a previous slow HRR underwent a second GXT within 3 months after their initial test. An additional eight subjects whose first GXT disclosed normal HRR were taken as a control group. Seven of 20 subjects demonstrated slow HRR (14 +/- 5 bpm) on the repeat test, and 13 subjects displayed normal HRR (29 +/- 5 bpm). Subjects with slow HRR in both GXT displayed significantly (P < 0.05) lower HF and LF fluctuations during recovery than those with normal HRR. Attenuated HRR after GXT, assessed by CWT, is indeed associated with abnormal vagal reactivation and prolonged sympathetic stimulation after termination of maximal exercise.