Since serum GH in normal subjects displays a circadian variation with a major and consistent surge after the onset of sleep we examined whether the time of GH administration in GH-deficient patients had any impact on its action. Eight GH-deficient patients all underwent 3 4-week study schedules in random order: 1) evening (2000 h) sc GH injections, 2) morning (0800 h) sc GH injections, and 3) no GH administration. At the end of each period the patients were admitted to hospital for 24-h measurements of hormones and metabolites. For comparison, 10 age- and sex-matched healthy untreated subjects were hospitalized once under identical conditions. Mean (+/- SE) GH availability, i.e. the area under the curve (AUC; micrograms per L/12 h) for 12 h after injection was significantly greater after evening injection than after morning injections [83.3 +/- 25.4 (evening) vs. 46.0 +/- 10.6 (morning); P less than 0.01]. This might be due to higher skin and sc temperatures when in bed. The 2000-0800 h AUC after evening injection was similar to the corresponding AUC in the reference group. Mean 24-h serum insulin-like growth factor-I levels (micrograms per L) were similar after evening (189.8 +/- 2) and morning (179.5 +/- 5.3) injections (P = 0.8), but the latter displayed a circadian variation suggesting that a steady state had not been reached. Both were significantly lower than the stable reference value (248.4 +/- 3.6 micrograms/L). Blood glucose profiles after morning and evening GH did not differ from that of the reference group, whereas blood glucose decreased when the patients received no GH (P less than 0.01). Daytime (0800-2400 h) insulin levels were increased after morning injections (P less than 0.05). Nighttime levels of lipid intermediates were below normal in the untreated state and after morning injection (P less than 0.05), whereas nighttime blood alanine tended to be above normal after morning GH injection (P = 0.08). Highly significant inverse relationships between circadian lipid intermediates and both blood alanine and lactate concentrations were observed in the reference group and in the patients after evening injections. These relationships disappeared after morning injections. We conclude that the metabolic effects of sc GH injections are clearly influenced by the time of administration and that the closest similarity to normal hormone and metabolite patterns and relationships is reached by GH injection in the evening in GH-deficient patients.