GH secretion was studied in 73 children with classical GH deficiency or GH neurosecretory dysfunction (GHND), intrinsic short stature, or normal stature. The GH-deficient group was defined by a peak GH secretory response below 10 ng/ml to all provocative tests (arginine, L-dopa, insulin hypoglycemia, and clonidine). GHND was defined by a mean serum 24-h GH concentration below 3 ng/ml, with a normal response (greater than or equal to 10 ng/ml) to provocative testing. Twenty-one GH-deficient children, 21 children with GHND, and 18 short control children underwent provocative GH testing and a 24-h study with GH sampling every 20 min. A group of 13 normal stature control children also underwent 24-h GH sampling. The mean stimulated peak serum GH level [4.7 +/- 0.6 (+/- SEM) ng/ml] in the GH-deficient group was significantly below that in the GHND (19.5 +/- 1.7 ng/ml) and short control groups (24.0 +/- 3.5 ng/ml; P less than 0.01). The mean 24-h serum GH concentration was reduced in GH-deficient (1.5 +/- 0.2 ng/ml) and GHND (2.0 +/- 0.1 ng/ml) children compared to those in short (5.6 +/- 0.5 ng/ml) and normal stature (5.8 +/- 0.8 ng/ml) control children (P less than 0.01). Peak GH concentrations after provocative testing correlated poorly with 24-h mean concentrations in GH-deficient, GHND, and short control children (r = 0.38, 0.23, and 0.41, respectively; P = NS for all groups). Mean serum GH concentrations from blood sampling intervals of 12 h (day/night; 0800-2000/2000-0800 h, respectively) or even 6 h (day; 0900-1500 h) were statistically different in GHND or GH-deficient groups compared to those in control children; however, there was significantly more overlap for individual children using the 6- and 12-h daytime intervals than for the 24-h data. Plasma somatomedin-C/insulin-like growth factor I correlated with mean 24-h GH concentration endogenous secretion (r = 0.7; P less than 0.001). These data suggest that provocative GH testing frequently does not correlate with endogenous GH secretion.