The effect of a 5-day fast on integrated, pulsatile, and periodic cortisol release was studied in 10 normal men by measuring serum cortisol concentrations every 20 min for 24 h before (day 0) and during the fifth day of fasting (day 5). Serum concentration profiles were analyzed for integrated cortisol release (area under the curve), pulsatile hormone release by an objective, statistically based pulse detection algorithm (cluster analysis), and periodic hormone release (circadian and ultradian rhythms) by Fourier expansion time series analysis. Urinary cortisol excretion per 24 h was measured in 5 men. The mean 24-h integrated serum cortisol concentration increased 1.7-fold during fasting (P = 0.0006). This increase resulted from a 2-fold increase in the serum cortisol concentrations between pulses (valley mean; P = 0.0004), an increase in the pulse height (P = 0.001), and an increase in pulse increment above baseline (P = 0.01). There were no changes in the number of pulses per 24 h, the interval between pulses, the width of the pulses, or the area of the pulses during fasting. Twenty-four-hour urinary cortisol excretion increased in all men, and the mean urinary cortisol (nanomoles per L)/creatinine clearance (milliliters per s) ratio increased from 119 on day 0 to 187 on day 5 (n = 5; P = 0.05). The pattern of periodic hormone release also changed during fasting; the mean (+/- SE) circadian rhythm (24-h) amplitude decreased from 160 +/- 14 nmol/L on day 0 to 102 +/- 105 nmol/L on day 5 (P = 0.06), and the amplitude of the 12-h rhythm increased from 68 +/- 11 to 99 +/- 11 nmol/L. There also were significant increases in the amplitudes of rhythms with periodicities of 8.1, 4.1, 2.4, 1.6, and 1.3 h (P = 0.02-0.008). Fasting in normal men results in distinct changes in the amount and pattern of pulsatile, circadian, and ultradian cortisol release.