To assess the degree of reproducibility of spontaneous GH secretion and pharmacological tests we studied 15 prepubertal children with short stature and abnormal growth rate. In all children, spontaneous overnight GH secretion was measured followed by a clonidine test in 8 children and an arginine test in the remaining 7. The same protocol was repeated a week later. Intra-individual variability of GH secretion in both physiological and pharmacological tests was expressed as the coefficient of variation (CV%). No significant difference was found between the first and second value of parameters of spontaneous GH secretion. Maximum spontaneous GH peak (MS) and mean 12-h GH concentration (MGH) correlated significantly (r = 0.78, p < 0.001). Mean CV% of all parameters of repeated GH profiles (around 30%) were lower than those of provocative tests (around 70%) (p < 0.05). No significant difference was found between CV% of clonidine and arginine tests. There was no correlation between MGH or MS and GH response to provocative test in the same child. We found a significant correlation between the log transformed maximum provocative GH response to the arginine test and the length of the time interval (in min) from the end of the last GH peak in the previous profile to the time zero of the provocative test (r = 0.60, p < 0.05). This relationship was not found for the clonidine test. We conclude that spontaneous GH secretion in children with short stature is more reproducible than stimulated GH response with a week's difference. Perhaps the higher variability of provocative GH secretion may be related to the state of the endogenous hypothalamic rhythm of both GHRH and somatostatin release at the time of the test.