OBJECTIVE: To evaluate a dose titration model for recombinant human GH substitution in adult patients with GH deficiency, aiming at normal plasma levels of IGF-I. DESIGN AND METHODS: Eighteen patients participated and a start dose of 0.17 mg GH/day was used except by two men who started with 0.33 mg/day. To demonstrate a clear GH effect the patients were first titrated, with steps of 0.17 mg GH/day every 6-8 weeks, to IGF-I levels in the upper range of age-adjusted reference values. The GH dose was then reduced 1 dose step and kept for a further 6 months. For comparison we investigated 17 healthy control subjects. RESULTS: Plasma IGF-I was increased after 2 weeks on the start dose and did not increase further for up to 8 weeks. Women had significantly lower GH sensitivity than men measured as net increment of IGF-I on the start dose of GH. GH sensitivity was not changed by age. The plasma IGF-I levels increased from 76.3+/-47.0 (s.d.) to 237+/-97 microg/l at the end of the study (P<0.001), and similar IGF-I levels were obtained in both sexes. The maintenance median GH dose was 0.33 mg/day in males and 0.83 mg/day in females (P=0.017). The GH dose correlated negatively with age in both sexes. Body weight, very low density triglycerides, lipoprotein(a) (Lp(a)), and fasting insulin increased, whereas insulin sensitivity index (QUICKI) decreased significantly. In comparison with the controls, the patients had lower fasting blood glucose, fasting insulin and Lp(a) levels at baseline, but these differences disappeared after GH substitution. The two groups had equal insulin sensitivity (QUICKI), but 2 h oral glucose tolerance test values of blood glucose and insulin were significantly higher in the patients at the end of the study. CONCLUSIONS: In conclusion our data suggest that the starting dose of GH substitution and the dose titration steps should be individualised according to GH sensitivity (gender) and the IGF-I level aimed for (age). The reduced insulin sensitivity induced by GH substitution could be viewed as a normalisation if compared with control subjects.