P revious observations in normal subjects have suggested that when 5-g glucose pulses (P) were given in the following sequence: before (P1) and 45 min after beginning a 300 mg/min glucose infusion (P2); during the 20th hr (P3) and 1 hr after the infusion was stopped (P4); the insulin responses were consistent with a simple two-pool model. One pool is a readily available small storage pool which is refilled by a second, larger, more slowly responding pool that controls basal and steady-state insulin output. The identical protocol was employed to evaluate the insulin responses in 13 nonobese diabetic subjects.Diabetics had basal insulin levels indistinguishable from normals (diabetics: 10.7±4; normals: 10.7±5, mean ±SD, μU/ml), but had significantly elevated basal glucose levels (diabetics: 161±27; normals: 88±7, mg/100 ml, P < 0.05). The mean early insulin response (3-5 min Δ IRI) after a 5 g glucose pulse (P1) was significantly diminished in diabetics (diabetics 6.4±9; normals: 32.5±14, μU/ml, P < 0.01) consistent with a defective storage pool output. The glucose disappearance rate, K_G, decreased in parallel with the early insulin response and the slope of the regression line between these two variables was virtually identical with that calculated from 16 normal subjects. Similar to normal subjects, during the short glucose infusion, the acute insulin response to P2 was diminished in diabetics (P < 0.02). In normal subjects after 20 hr of infusion, the rapid insulin responses to P3 are restored to the preinfusion P1 values, and 1 hr after the infusion was stopped, the responses to P4 are increased twofold (P < 0.001). Diabetics, however, demonstrated decreased early responses to P3 (P < 0.001) and no increased response to P4.In contrast to the diminished acute insulin responses to glucose pulses, diabetics have steady-state insulin levels after 20 hr of glucose infusion similar to those of normal subjects (diabetics: 25.7±13; normals: 32.5±14, μU/ml). Thus both basal and steady-state insulin levels of diabetics were comparable with those of normal subjects, which suggest that although the rapid insulin response from the storage pool output is defective in diabetics, the more slowly responding pool is intact.