OBJECTIVE A number of studies of the effect of GH replacement therapy in adult patients with GH deficiency have been published, but the definition of GH deficiency has varied considerably. In order to define severe GH deficiency more critically we have determined GH status in the context of gonadotrophin, ACTH and TSH secretion in adult patients with pituitary disease. METHODS Analysis of peak GH response to an insulin tolerance test performed during comprehensive assessment of pituitary function. METHODS One hundred and ninety non-acromegalic patients (96 male) with pituitary disease whose ages ranged from 16 to 72 (mean 39.4) years. METHODS The patients were divided into four groups according to the number of anterior pituitary hormone deficiencies demonstrated; isolated GH deficiency (GHD0), or GH deficiency plus an additional one, two or three pituitary hormone deficits (GHD1, GHD2, GHD3). RESULTS The four groups were matched for age and blood glucose nadir during the ITT. The median (interquartile range) GH peaks were GHD0, 10.0 (5.4-16); GHD1, 4.0 (2.7-7.7); GHD2, 2.0 (1-2.9); GHD3, 1.8 (1-3.2) mU/l. There was a significant downward trend in the medians (P < 0.0001). The differences between GHD0 and GHD1, and GHD1 and GHD2, were highly significant (P < 0.0001); however, there was no difference between GHD2 and GHD3. Ninety-one per cent of patients in combined groups GHD2 and GHD3, 55% in GHD1 and 24% in GHD0 had a peak GH < 5 mU/l. CONCLUSIONS Our study has shown that GH deficiency is variable according to the degree of hypopituitarism present and that the greater the number of pituitary hormone deficits the more severe the GH deficiency. These observations will help to clarify the diagnosis of GH deficiency in adult life.