Glucocorticoid excess may be associated with poor growth despite normal levels of GH and adequate nutrition. Steroid-induced growth failure could be mediated by defective generation and/or action of somatomedins. To probe potential mechanisms, we examined the effect of corticosteroid administration on net somatomedin activity, immunoreactive somatomedin-C, and separated biologically active somatomedins and somatomedin inhibitors. Twelve children receiving alternate day steroid therapy had circulating somatomedin activity measured by porcine cartilage bioassay. Somatomedin activity fell 6 h after steroids [from 1.02 +/- 0.09 (+/- SEM) to 0.35 +/- 0.07 U/ml; P less than 0.001] and then rose toward normal. No significant change in somatomedin activity occurred during the day off therapy. Further studies were conducted in normal subjects given a single 60-mg dose of prednisone. Six hours after prednisone, somatomedin activity (rat cartilage bioassay) decreased by 46% (P less than 0.01), yet somatomedin-C did not change. To pursue this discrepancy, serum was fractionated on Sephadex G-50, pH 2.4, and separated somatomedin and somatomedin inhibitory bioactivity was measured. Biologically active somatomedins (Kav, 0.50-0.63) were comparable before and after prednisone treatment, as was inhibitory activity found at Kav 0.13-0.25. In contrast, somatomedin inhibitory activity at Kav 0.25-0.38 doubled (111 +/- 8% inhibition of somatomedin action vs. 54 +/- 11%; P less than 0.005) after prednisone therapy. The somatomedin inhibitor in these fractions blunted serum stimulation of sulfate, thymidine, and uridine uptake by test cartilage. These inhibitory effects could not be attributed to direct steroid action, as levels were less than 2 micrograms/dl in inhibitory fractions and addition of cortisol and prednisolone to the bioassay system failed to decrease somatomedin activity. We conclude that glucocorticoid administration is followed by an increase in circulating somatomedin inhibitors. Such inhibitors may explain the steroid-induced fall in net somatomedin activity and contribute to impaired growth.