After completion of treatment with growth hormone (GH) 19 patients with isolated 'idiopathic' GH deficiency and 15 with post-irradiation GH deficiency underwent retesting of GH secretion with an insulin tolerance test or an arginine stimulation test, or both. Patients with post-irradiation GH deficiency comprised 13 patients with central nervous system tumours distant from the hypothalamo-pituitary axis and two with acute lymphoblastic leukaemia, who had received cranial or craniospinal irradiation. All 15 patients with post-irradiation GH deficiency remained GH deficient (peak GH response less than 7 mU/l (n = 10) and 7-15 mU/l (n = 5)). Of the 19 retested patients with idiopathic GH deficiency, however, five (26%) had peak GH responses of greater than 15 mU/l (regarded now as transient or false idiopathic GH deficiency) and were indistinguishable from the remainder (permanent or true idiopathic GH deficiency, peak GH responses less than 7 mU/l (n = 12) and 7-15 mU/l (n = 2)), by pretreatment anthropometry and post-treatment height standard deviation score, but had a lower first year height velocity (mean (SD) velocity 5.6 (0.5) cm/year for false idiopathic deficiency v 8.7 (1.75) cm/year for true idiopathic deficiency, p less than 0.01) and height increment on treatment (mean (SD) increment 2.2 (1.5) cm/year for false idiopathic deficiency v 5.2 (2.3) cm/year for true idiopathic deficiency, p less than 0.05). By current practices two patients with false idiopathic deficiency may have been distinguished by sex steroid priming. Thus post-irradiation GH deficiency seems to be permanent, but errors in diagnosis in idiopathic GH deficiency are common.