The growth hormone (GH) reserve of 15 short children was evaluated with the levodopa-propranolol test (DPT) and the sequential arginine-insulin test (AIT). Four patients failed to respond to both tests and were classified as hyposomatotropic. In the other 11 children, the mean GH peak response to the DPT was significantly higher than that to the AIT, mainly because five subjects who had a normal response to the DPT failed to respond to the AIT. These children had a generally poor yearly growth increment prior to testing associated in three with an obvious emotional problem, and were found at follow-up to have resumed a normal growth pattern. These data confirm the effectiveness of the DPT as a test of GH reserve. Although hypoglycemia can occur occasionally during the test, this procedure is safer and easier to perform than the widely used AIT. Finally, the DPT seems to detect a category of children who have a temporary growth failure and nonresponse to the usual GH tests but who are not hyposomatotropic and consequently do not require human GH.