Plasma insulin is pulsatile and reflects oscillatory insulin secretion from pancreatic islets. Although both islet Ca(2+) and metabolism oscillate, there is disagreement over their interrelationship, and whether they can be dissociated. In some models of islet oscillations, Ca(2+) must oscillate for metabolic oscillations to occur, whereas in others metabolic oscillations can occur without Ca(2+) oscillations. We used NAD(P)H fluorescence to assay oscillatory metabolism in mouse islets stimulated by 11.1 mM glucose. After abolishing Ca(2+) oscillations with 200 microM diazoxide, we observed that oscillations in NAD(P)H persisted in 34% of islets (n = 101). In the remainder of the islets (66%) both Ca(2+) and NAD(P)H oscillations were eliminated by diazoxide. However, in most of these islets NAD(P)H oscillations could be restored and amplified by raising extracellular KCl, which elevated the intracellular Ca(2+) level but did not restore Ca(2+) oscillations. Comparatively, we examined islets from ATP-sensitive K(+) (K(ATP)) channel-deficient SUR1(-/-) mice. Again NAD(P)H oscillations were evident even though Ca(2+) and membrane potential oscillations were abolished. These observations are predicted by the dual oscillator model, in which intrinsic metabolic oscillations and Ca(2+) feedback both contribute to the oscillatory islet behavior, but argue against other models that depend on Ca(2+) oscillations for metabolic oscillations to occur.