Random plasma somatomedin-C (SM-C) levels were measured in 143 children and adolescents with growth at or below the 5th percentile. 124 patients had short stature due to constitutional delay or genetic predisposition and 19 patients were growth hormone deficient (GHD). When analyzed according to bone age, mean somatomedin-C levels rose gradually with increasing bone age in non-GHD patients, while there was no change with increasing bone age in GHD patients. All patients with SM-C levels < 0.5 U/ml underwent growth hormone testing to rule out GHD. This criterion resulted in 52% accuracy in identifying non-GHD patients and 100% accuracy in identifying GHD patients. Comparison of the SM-C test with the exercise GH screening test demonstrated the advantage of greater convenience for the SM-C test, and comparable cost effectiveness despite a lower specificity associated with the SM-C test. SM-C levels were not predictive of peak growth hormone response nor could they be used to differentiate constitutional delay of growth and maturation from genetic short stature.