Glucocorticoids are used in an attempt to reduce brain edema secondary to head injury. Nevertheless, their usefulness remains uncertain and contradictory. In a randomized study of 24 children with severe head injury, urinary free cortisol was measured by radioimmunoassay. Twelve patients (group 1) received dexamethasone and 12 (group 2) did not. All patients were treated with a standardized regimen. In group 1 there was complete suppression of endogenous cortisol production. In group 2 free cortisol was up to 20-fold higher than under basal conditions and reached maximum values on days 1-3. Since the excretion of cortisol in urine reflects the production rate closely and is not influenced by liver function and barbiturates, the results in group 2 show that the endogenous production of steroids is an adequate reaction to severe head injury. Exogenous glucocorticoids are thus unlikely to have any more beneficial effects than endogenous cortisol.