In order to investigate the relationship between pituitary appearance and the diagnosis of growth hormone deficiency (GHD), we have assessed magnetic resonance imaging (MRI) scans and GH status during provocation tests in 110 patients (78 males; median age 9.8, range 0.1-20 yr), evaluated for possible GH disorders. On the basis of pituitary function tests, patients were divided into GH deficient (GH peak < 15 mIU/l [5.8 ng/ml]) (n = 82) or GH sufficient (GH peak > 15 mIU/l) (n = 28). The former were further divided into those with multiple hormone deficits (MPHD) (n = 19) or isolated GHD - severe IGHD (peak GH < 8 mIU/l [3.1 ng/ml]) or partial IGHD (8-15 mIU/l). The appearance of the hypothalamic-pituitary (H-P) axis was classified as: (1) normal, (2) isolated hypoplastic stalk (HPS), (3) isolated hypoplastic anterior lobe (HPAL) (PHT SDS < -2.0), (4) HPS + HPAL or (5) ectopic posterior lobe (EPL). The last two were considered severe abnormalities. PHT SDS (mean +/- SD -2.0 +/- 2.2) was correlated to log peak GH levels in the whole group (r = 0.45; p < 0.0001) and in the GHD group (r = 0.39; p < 0.0001). Sixty-five out of 82 in the GHD group had a H-P axis abnormality (45 severe abnormalities), while 13 out of the 28 patients in the GH sufficient group also had an abnormality (3 severe, but none with an EPL). All patients with MPHD had a MRI abnormality, most commonly an EPL (79%). Thus the presence of any MRI abnormality as a marker for GHD would generate a sensitivity of 79%, but a specificity of only 54%, indicating that this could not be used to confirm GHD. However, the presence of either an EPL or HPS + HPAL on MRI is highly specific (100% and 89% respectively) and predictive of GHD (positive predictive value 100% and 79% respectively), indicating that these abnormalities provide confirmation of the diagnosis. We recommend that if clinical, auxological and biochemical data indicate a diagnosis of GHD, then a MRI scan should be undertaken to define the pituitary anatomy and to help confirm the diagnosis.