Objective - To determine the effect of nocturnal hypoglycaemia on sleep architecture in adolescents with insulin dependent diabetes mellitus (IDDM). Design - 20 adolescents with IDDM (mean age 12.8 years, mean glycated haemoglobin (HbA(1c) 8.9%) were studied on one night. Plasma glucose was measured every 30 minutes and cortisol and growth hormone levels every 60 minutes. Sleep was recorded using standard polysomnographic montages, and sleep architecture was analysed for total sleep time, stages 1-4, rapid eye movement, fragmentation, and arousals. Results - Six subjects (30%) became hypoglycaemic (five subjects < 2.5 mmol/l), with one being symptomatic. There were no differences in age, HbA(1c), duration of diabetes, or insulin regimen between hypoglycaemic and non-hypoglycaemic subjects. Hypoglycaemia was not predicted by glucose measurements before bed. There was no detectable rise in plasma cortisol or growth hormone concentrations during hypoglycaemia. Sleep architecture was not disturbed by nocturnal hypoglycaemia with no differences found in sleep stages, fragmentation, or arousals. Conclusions - Nocturnal hypoglycaemia is a common and usually asymptomatic complication of treatment in adolescents with IDDM. Moderate hypoglycaemia has not been shown to affect sleep architecture adversely. These findings are consistent with, and may explain, the observation that severe hypoglycaemia, with consequent seizure activity, is more common at night than during the day. Counter-regulatory hormone responses to nocturnal hypoglycaemia may be less marked than with similar degrees of diurnal hypoglycaemia.