The pathogenesis of the abnormal metabolic state in patients with non-insulin-dependent diabetes (NIDDM) is controversial. Even the term NIDDM stirs controversy because of the easily drawn inference that individuals with this form of diabetes do not need insulin treatment. Yet many patients with NIDDM are treated with insulin; some even develop hyperosmolar coma if not given insulin. Ketoacidosis, however, is very infrequent in this syndrome, implying that these patients are not dependent on insulin treatment to prevent mass mobilization of fatty acids and ketone bodies. The phrase noninsulin-dependent is therefore appropriate when used in this restricted fashion but inappropriate when used to imply adequacy of insulin secretion. The evaluation of the adequacy of islet function in this syndrome has been complex, since there is no standard of insulin output that can be defined for normal islets without specifying the physiologic setting under which the assessment has been made. For example, individuals with normal glucose levels but variable degrees of obesity have widely varying insulin secretion rates. Thus, the choice of controls is critical when comparing islet function in NIDDM to normal. In addition, the efficiency of the B-cell response to a challenge (e.g., oral glucose tolerance test) can markedly influence the magnitude of the stimulatory glucose level during the period of testing. For example, a subject with some impairment of insulin output will tend to become more hyperglycemic during the test. The hyperglycemia may then stimulate more insulin secretion so that the overall insulin output may appear equal to or even greater than that of a normal individual. In such a closed-loop system, strict control of input variables is necessary to evaluate whether or not insulin secretion is normal. As will be discussed, control of glucose level and other variables is seldom accomplished in dynamic glucose tolerance tests. As will be presented in this review, the development of appropriately controlled studies of islet function has provided convincing evidence that islet B-cell function is abnormal in patients with NIDDM. Since these studies are based on an understanding of normal islet function, normal islet B-cell physiology is discussed before pathophysiology. Finally, the implications of this analysis for the treatment of NIDDM with diet, hypoglycemic sulfonylureas, and insulin will be discussed.