Little information is available regarding the optimal timing of exercise in insulin-dependent diabetes mellitus (IDDM) patients. In this study, six IDDM patients receiving ultralente-based intensive insulin therapy were studied during 30 min of exercise (~60% Vo(2max)), before breakfast, and at 1600. On two other occasions, they were studied at rest. Plasma glucose increased from 6.7 ± 0.4 to 9.1 ± 0.4 mM during morning exercise (P <0.01). In contrast, mean plasma glucose did not change during afternoon exercise (Δ = 0.3 ± 0.5 mM, NS); however, there was a 0.3- to 1.0-mM decrease in three subjects. The observed difference in the glycemic response to exercise could not be explained on the basis of changes in plasma glucagon, growth hormone, norepinephrine, or epinephrine. Plasma cortisol was higher (P <0.02) in the morning than in the afternoon, and plasma free-insulin concentrations were lower (P <0.05). These data indicate that the risk of exercise-induced hypoglycemia is lowest before breakfast. The reason for the divergent glycemic responses to exercise is not entirely clear but may be related to the observed differences in free-insulin concentrations. Because of the lower risk of hypoglycemia, our results suggest prebreakfast exercise may be preferable for some IDDM patients receiving intensive insulin therapy. Whether these findings are relevant to patients receiving other types of insulin therapy will require further investigation.