We evaluated the role of insulin availability in the induction therapy of non-insulin dependent diabetes mellitus (NIDDM). Plasma glucose (BG) and insulin (IRI) response to glucose loading (OGTT) was investigated before and after placebo and phenobarbitone (PB) therapy in patients with glucose intolerance and NIDDM treated with diet only, sulphonylureas (SU) plus metformin (M) or insulin. The antipyrine test was used to reflect the liver mixed function oxidase system. Therapy with PB, but not placebo, reduced fasting IRI and BG in subjects with glucose intolerance and improved the glucose tolerance, insulin response to glucose and antipyrine metabolism. The effects of PB on patients with NIDDM were dependent on the insulin availability and duration of the disease. Best responses were seen in hyperinsulinemic patients at the early phase of the disease. They had lowered fasting BG and IRI values, improved glucose tolerance, insulin response to OGTT and antipyrine metabolism after PB therapy. The SU plus M treated patients responded beneficially if they had high fasting and postglucose IRI values and were non-responders if they had relative insulin deficiency. Antipyrine metabolism improved among the responders and non-responders. The hyperglycemic patients treated with insulin showed improved glucose metabolism, an improved C-peptide response and antipyrine metabolism. The subjects could be classified into responders and non-responders by calculating the ratio of areas above the fasting level curve of insulin and glucose (sigma delta I-OGTT/sigma delta G-OGTT). These values for the former were 0.2-0.4 and the latter 0.03-0.04, as compared to healthy volunteers (1.0) and subjects with glucose intolerance (1.4). A PB type inducer improves insulin sensitivity but does not alter its production or secretion. The outcome of glucose metabolism is therefore dependent on insulin availability.