The effect of thiazide diuretics on the glucose tolerance of seven normal men in whom potassium loss was prevented with supplementation was studied using the glucose clamp technique. An initial control 2-h hyperglycemic clamp was performed to create a square wave of hyperglycemia 125 mg/dl above basal. At 1 h, 40 g glucose/m2 body surface area was ingested. Serial insulin (IRI) and gastric inhibitory polypeptide (GIP) levels were measured as well as the level of glucose infusion necessary to maintain the stable hyperglycemic level. After the initial study, subjects were placed on a 10-day course of 100 mg hydrochlorothiazide and 80 meq potassium per day. Subjects were monitored for dietary potassium intake, urinary potassium, and sodium losses, and the replacement of potassium adjusted accordingly. A repeat glucose clamp was done on day 10. When potassium losses were prevented, thiazides induced no alterations in glucose tolerance, beta-cell sensitivity to glucose, GIP-cell sensitivity to glucose, beta-cell sensitivity to GIP, or tissue sensitivity to insulin. Two control studies in which hypokalemia was allowed to ensue after hydrochlorothiazide ingestion revealed a diminution in glucose tolerance, a consequence of diminished pancreatic beta-cell response to glucose. We conclude that the thiazide effect on glucose tolerance is a consequence of the resultant hypokalemia that the diuretic may create.