Diurnal variations in cortisol response to exercise and their relation to the events of the day were assessed by comparing daily cortisol patterns on the control day and on days when exercise was performed at different times. A remarkable midday cortisol peak coincided with the noon meal, but cortisol levels were irregularly affected by an identical meal in the evening. However, the exercise produced equal increases when performed during quiescent periods (i.e. without any secretory peaks at 1000, 1430, 1700, and 2130 h), but peak levels for exercise at 2130 h were significantly lower because of the lower basal levels in the evening. When the same exercise was performed at 1300 h (i.e. coinciding with the postprandial peak), only a brief leveling-off interrupted the decline in cortisol levels. The meal-related evening peak, if any, provoked a similar decrease in response to exercise performed at 2000 h. Similarly, the midday peak itself was reduced by a prior exercise-induced cortisol rise. These results show that the daily cortisol pattern results from the interactions between the meal-related peaks, especially the major midday cortisol peak, and the exercise-induced increases, both of which inhibit the response to subsequent stimulation. The identical responses to exercise at the different quiescent periods tested, despite a general downward trend in basal cortisol levels, establish the primacy of such feedback mechanisms over those responsible for the circadian rhythm.