Between January 1985 and May 1986, following completion of a pilot study, a main study concerning the possible association between Reye's syndrome and salicylates was conducted. Twenty-seven patients with stage II or deeper Reye's syndrome whose diagnoses were confirmed by an expert panel and who had appropriate antecedent illnesses (chickenpox, respiratory illness, or gastrointestinal illness) prior to the onset of Reye's syndrome were compared with 140 controls matched for age, race (black or not black), and type and timing of onset of antecedent illness. Controls were selected from the same hospital, emergency room, or school as case-patients or were identified by random-digit telephone dialing. As in the pilot study, a strong statistical association with ingestion of salicylates during the antecedent illness and prior to the onset of Reye's syndrome was observed (odds ratio, 40; lower 95% confidence limit, 5.8). Analysis of the independent risk of aspirin and nonaspirin salicylates revealed a significant association with aspirin (odds ratio, 26; lower 95% confidence limit, 6.4); the independent risk of nonaspirin salicylates could not be assessed because only two cases were not exposed to aspirin. Assessment of epidemiologic issues of concern, including case-control differences in the severity of the antecedent illness, did not explain the high odds ratios that were observed. The high percentage of patients with Reye's syndrome exposed to salicylates (≥90%) in this and prior studies suggests that, though the reported incidence of Reye's syndrome has declined in recent years, concomitant with a decline in salicylate use among children, a majority of Reye's syndrome cases may be attributable to salicylate use.