The aim of this study was to see if the greater effect of insulin on hepatic glucose output when insulin is given using 13-min pulses in man remains when the same amount of insulin is delivered using 26-min pulses. The study was performed on nine male healthy volunteers submitted to a 325 min glucose-controlled glucose iv infusion using the Biostator. The endogenous secretion of pancreatic hormones was inhibited by somatostatin. Three experiments were performed in each subject on different days and in random order. In all cases glucagon was replaced (58 ng min-1). The amounts of insulin infused were identical in all instances and were 0.2 mU kg-1 min-1 (continuous), 1.3 mU kg-1 min-1, 2 min on and 11 min off (13-min pulses) or 2.6 mU kg-1 min-1, 2 min on and 24 min off (26-min pulses). Blood glucose levels and glucose infusion rate were monitored continuously by the Biostator, and classic methodology using D-[3-3H] glucose infusion allowed to study glucose turnover. When compared with continuous insulin, 13-min insulin pulses induced a significantly greater inhibition of endogenous glucose production. This effect disappeared when insulin was delivered in 26-min pulses. We conclude that, in man, an adequate pulse frequency is required to allow the appearance of the greater inhibition of pulsatile insulin on endogenous glucose production.