To test the hypothesis that nocturnal hypoglycemia causes fasting hyperglycemia (the Somogyi phenomenon) in patients with insulin-dependent diabetes mellitus, we studied 10 patients, who were on their usual therapeutic regimens, from 10 p.m. through 8 a.m. on three nights. On the first night, only a control procedure was performed (blood sampling only); on the second night, hypoglycemia was prevented (by intravenous glucose infusion, if necessary, to keep plasma glucose levels above 100 mg per deciliter [5.6 mmol per liter]); and on the third night, hypoglycemia was induced (by stepped intravenous insulin infusions between midnight and 4 a.m. to keep plasma glucose levels below 50 mg per deciliter [2.8 mmol per liter]). After nocturnal hypoglycemia was induced (36 +/- 2 mg per deciliter [2.0 +/- 0.1 mmol per liter] [mean +/- SE] from 2 to 4:30 a.m.), 8 a.m. plasma glucose concentrations (113 +/- 18 mg per deciliter [6.3 +/- 1.0 mmol per liter]) were not higher than values obtained after hypoglycemia was prevented (182 +/- 14 mg per deciliter [10.1 +/- 0.8 mmol per liter]) or those obtained after blood sampling only (149 +/- 20 mg per deciliter [8.3 +/- 1.1 mmol per liter]). Indeed, regression analysis of data obtained on the control night indicated that the 8 a.m. plasma glucose concentration was directly related to the nocturnal glucose nadir (r = 0.761, P = 0.011). None of the patients was awakened by hypoglycemia. Scores for symptoms of hypoglycemia, which were determined at 8 a.m., did not differ significantly among the three studies. We conclude that asymptomatic nocturnal hypoglycemia does not appear to cause clinically important fasting hyperglycemia in patients with insulin-dependent diabetes mellitus on their usual therapeutic regimens.