Summary. Twenty diabetic patients, whose hyperglycaemia had been shown to fail to respond to at least one month's dietary treatment, completed a crossover study in order to: 1) compare the effectiveness of two sulphonylureas, chlorpropamide and glibenclamide, and 2) study the effects of sulphonylureas on insulin secretion and on biochemical indices of glucose intolerance. Fasting blood glucose fell on active treatment from 10.7 _-2 0.6 (mean _.+ SEM) to 6.6 + 0.7 mmol/l and rose again to 10.6 z 0.7 after 4 months placebo. A second period of 4 months sulphony/urea therapy resulted in a comparable fatl in blood glucose (to 6.9 + 0.7 mmol/1) and 'a similar relapse was seen after the second placebo period (to 10.5 ___ 0.9 mmol/i). Glucose tolerance and associated insulin secretion improved markedly on active treatment, with ketone bodies, non-esterified fatty acids, and glycerol falling to within the reference range. Sulphonylurea therapy was associated with a small but significant increase in the fasting insulin level. These effects were nearly all reversed 4 months after withdrawal of the sulphonylureas. No marked changes were found in growth hormone, lactate, pyruvate, lactate/pyruvate ratio or fasting cholesterol, trigtycerides and lipoproteins. On a weight basis, glibenclamide was 26 times more potent than chlorpropamide and, in the doses used in this study, their biochemical effects were indistinguishable. The effects of these two sulphonylureas seem most likely to be mediated by a direct stimulation of insulin secretion by the B-cell.