Livers from fed rats (180-240 g) were perfused noncyclically with a hemoglobin-free medium in vitro to determine whether vasoactive intestinal peptide (VIP) increases hepatic glucose production through a cAMP- or a Ca(2+)-dependent mechanism. Glucose output did not increase, but cAMP increased maximally during 10(-9) M VIP infusion. When VIP was perfused at 10(-8) M or more, glucose output increased dose dependently, whereas cAMP increased only a little during the VIP infusion, but increased greatly after the infusion. When Ca2+ was excluded from the perfusate, glucose output produced by 10(-8)-10(-7) M VIP was only 40% of that observed in the Ca(2+)-containing perfusion, and the increase in cAMP was abolished almost completely. By adding 10(-7) M A23187 for 10 min during the infusion of 10(-9) M VIP, cAMP, which increased with VIP alone, decreased during the A23187 infusion and increased again after the cessation of the A23187 infusion, whereas glucose output increased during the A23187 infusion. These results were similar to those observed with higher concentrations of VIP. When 10(-4) M isobutylmethylxanthine and 10(-8) M VIP were infused concurrently, cAMP increased rapidly during the infusion and decreased after the infusion. In conclusion, 1) glycogenolysis is produced by VIP through a Ca(2+)-dependent mechanism, rather than a cAMP-dependent one; and 2) the restriction of cAMP accumulation during the infusion of high concentrations of VIP is caused by Ca(2+)-induced phosphodiesterase activation.