Forty-eight narcoleptic patients were randomly selected from a group of 197 narcoleptic outpatients, and a 50-g oral glucose tolerance test (OGTT) was performed. Blood glucose levels were determined by oximetry before and 30, 60, and 120 min after the glucose administration. Serum insulin concentrations (IRI) were determined by enzyme immunoassay, and an insulinogenic index (II) was calculated. (II = the ratio of IRI increment to blood glucose increment in 30 min following glucose load). From the results of the OGTT, six patients with definite diabetes mellitus (DM) were identified (12.5%, 4 men and 2 women) according to World Health Organization criteria (1980). The II of the DM patients was significantly low, with an average of 0.13 +/- 0.03. From recent data in the literature on the prevalence of DM in the general Japanese adult population (1.75-5.5%), it was concluded that the frequency of DM among our randomly selected narcoleptic patients was significantly increased. The positive results for diabetes in our six narcoleptic patients could not be attributed to obesity, because there was no difference between the obesity indices of DM and nonDM narcoleptic patients, nor was there a difference between the frequency of DM in obese and nonobese patients.