The diagnosis of GH deficiency (GHD) in the elderly is based at present on the peak GH concentration during a stimulation test. We have now evaluated the performance of urinary GH (uGH), urinary insulin-like growth factor I (uIGF-I), and urinary IGF-binding protein-3 (uIGFBP-3) in the diagnosis of GHD in this group. Twenty GHD elderly patients with a history of pituitary disease and a peak GH response to arginine stimulation of less than 3 ng/mL (15 men and 5 women; age, 61.1-83.4 yr) and 19 controls (12 men and 7 women; age, 60.8-87.5 yr) were studied. GH secretion was assessed by 24-h profile and expressed as the area under the curve (AUCGH). Serum (s) IGF-I and sIGFBP-3 were measured in a single morning, fasted sample. Urinary GH, uIGF-I, and uIGFBP-3 were measured in a 24-h urine sample collected over the same interval as the GH profile, and results were expressed as total amount excreted in 24 h (tuGH24, nanograms; tuIGF-I24, nanograms; tuIGFBP-3(24), micrograms). Data are presented as the mean +/- SD, except for AUCGH, tuGH24, and tuIGFBP-3(24), which are presented as the geometric mean (-1, +1 tolerance factor). AUCGH, sIGF-I, and sIGFBP-3 were significantly lower in GHD subjects than in controls. Total uGH24 was lower in GHD subjects, but tuIGF-I24 and tuIGFBP-3(24) excretion were not different in the two groups. AUCGH provided the best separation between GHD and control subjects, whereas there was substantial overlap for sIGF-I, sIGFBP-3, and tuGH24. In both groups sIGF-I was correlated to sIGFBP-3 (GHD, r = 0.75; controls, r = 0.65; both P < 0.01), whereas tuIGF-I24 was not correlated to tuIGFBP-3(24) in either group. Moreover, tuIGF-I24 and tuIGFBP-3(24) were not related to their respective serum concentrations in either group. Total uGH24 was correlated with AUCGH only in controls (r = 0.54; P < 0.05). These data demonstrate that urinary GH and urinary and serum IGF-I and IGFBP-3 are not suitable diagnostic markers for GHD in elderly subjects.