Recent evidence suggests that variations in cortisol activity within the physiological range contribute to associations between multiple cardiovascular risk factors. Plasma cortisol measurements during a glucose tolerance test differ in men with hypertension, insulin resistance, and glucose intolerance, but it is unclear whether this reflects altered responses of cortisol to glucose, altered circadian rhythm, or altered habituation to multiple sampling. We performed a single-blind randomized cross-over study comparing 75 g oral glucose with placebo in 39 fasted men (22 glucose intolerant and 17 controls) aged 68-77 yr. In all subjects, plasma cortisol fell during the glucose tolerance test. Subjects with glucose intolerance had significantly higher plasma cortisol following placebo (P = 0.001), suggesting an altered circadian rhythm. Treatment with an oral glucose load blunted the circadian fall in plasma cortisol (P = 0.002), but this response was no different in controls or glucose intolerant subjects. In addition, 0900 h plasma cortisol was higher in the first study phase in controls (P = 0.01) but not in glucose-intolerant subjects (P = 0.18), who showed a lack of habituation to repeated plasma measurements. These data support the hypothesis that alterations in central regulation of the hypothalamic-pituitary-adrenal axis may be important in glucose intolerance.