During a 22-year period 51 cases of Stevens-Johnson syndrome (SJS) occurred in infants and children who ranged in age from 3 months to 14 years. Fifty-six percent of patients had an antecedent upper respiratory tract infection or non-specific viral infection, and 67% had received a prescription medication in the 3 weeks before onset of SJS. Nineteen patients (37%) were treated with adrenocorticosteroid medication during their hospitalization. Age, sex, duration of illness, body temperature on admission and history of antecedent medication were similar for the steroid- and non-steroid-treated patients. Rates of infection and overall complications were significantly greater in steroid-treated patients than in those treated symptomatically. These observations suggest that steroid drugs should not be used for treatment of SJS in infants and children.