The causes of growth retardation of children with thalassaemia major are multifactorial. We studied the GH response to provocation by clonidine and glucagon, measured the circulating concentrations of insulin, IGF-I, IGF-binding protein-3 (IGFBP-3) and ferritin, and evaluated IGF-I generation after a single dose of GH (0.1 mg/kg per dose) in 15 prepubertal patients with thalassaemia, 15 age-matched children with constitutional short stature (CSS) (height standard deviation score less than ¹2, with normal GH response to provocation) and 11 children with isolated GH deficiency (GHD). Children with thalassaemia had significantly lower peak GH response to provocation by clonidine and glucagon (6:2 Ϯ 2:3 and 6:8 Ϯ 2:1 mg/l respectively) than the CSS group (18:6 Ϯ 2:7 and 16:7 Ϯ 3:7 mg/l respectively). They had significantly decreased circulating concentrations of IGF-I and IGFBP-3 (47:5 Ϯ 19 ng/ml and 1:2 Ϯ 0:27 mg/l respectively) compared with those with CSS (153 Ϯ 42 ng/ ml and 2:06 Ϯ 0:37 mg/l respectively), but the IGF-I and IGFBP-3 concentrations were not different from those with GHD (56 Ϯ 25 ng/ml and 1:1 Ϯ 0:32 mg/l respectively). These data demonstrate that the GH-IGF-I-IGFBP-3 axis in thalassaemic children is defective. Serum ferritin concentration correlated significantly with GH peak response to provocation (r ¼ ¹0:36, P < 0:05) and circulating IGF-I (r ¼ ¹0:47, P < 0:01) and IGFBP-3 (r ¼ ¹0:42, P < 0:01) concentrations. In the IGF-I generation test, after GH injection, the thalassaemic children had significantly lower IGF-I and IGFBP-3 levels 86:7 Ϯ 11:2 ng/ml and 2:05 Ϯ 0:51 mg/l respectively) than those in the CSS group (226 Ϯ 45:4 ng/ ml and 2:8 Ϯ 0:43 mg/l respectively). The IGF-I response was significantly higher in children with GHD (158 Ϯ 50 ng/ml) than in thalassaemic children. Six short (height standard deviation score less than ¹2) thalassaemic children who had defective GH response to provocation (<10 mg/l), all the children with GHD and eight short normal children (CSS) were treated for 1 year with human GH (18 units/m 2 per week divided into daily s.c. doses). After 1 year of GH therapy there was a marked acceleration of growth velocity in both thalassaemic children (from 3:8 Ϯ 0:6 cm/year to 7:2 Ϯ 0:8 cm/year) and controls. However, the linear acceleration of growth velocity on GH therapy was significantly slower in thalassaemic children (3:3 Ϯ 0:3 cm/year increment) compared with those with CSS (5:3 Ϯ 0:4 cm/year increment) and GHD (6:9 Ϯ 1:2 cm/year increment) (P < 0:05). Their circulating IGF-I concentration (105 Ϯ 36 ng/ml) was significantly lower than those for CSS (246 Ϯ 58 ng/ml) and GHD (189 Ϯ 52 ng/ml) after 1 year of GH therapy. These data prove that some children with b-thalassaemia major have a defective GH-IGF-I-IGFBP-3 axis and suggest the presence of partial resistance to GH.