Serum salicylate concentration was measured at admission in 130 children with liver-biopsy-confirmed Reye's disease. Mean serum salicylate was 12.3 mg/dl and mean salicylate concentrations by neurological grade (Lovejoy) were: stage I, 12, stage II, 13, stage III, 11, stage IV, 13, and stage V, 13 mg/dl. However, mean serum salicylate (15 mg/dl) at admission in 21 patients who died or had serious neurological deficits was significantly higher than that in 103 patients who survived without neurological sequelae (10 mg/dl). Serum salicylate in a group of 27 age-matched, community-matched control children collected consecutively over the period 1978-80 was less than 2 mg/dl, and children with varicella or influenza had salicylate concentrations indistinguishable from apparently well classmates or siblings. It is impossible to determine from this data whether salicylates are involved in the aetiology of or in determining the outcome of Reye's disease. Increased concentrations of salicylates at admission could be the result of excessive dosage because of a greater severity of the prodromal illness, or to diminished excretion because of impaired hepatic metabolism. It seems likely that serum salicylate concentrations entered the toxic range in many patients with Reye's disease before they presented for treatment. Most had been vomiting and had diminished oral intake for 33-55 h before hospital admission. Since the average number of hours from the beginning of vomiting to admission was no different in non-comatose and comatose cases, the time at which salicylate concentration was measured in relation to the last dose was probably similar in the two groups and therefore does not account for the higher levels in children with poor outcome. Salicylates are mitochondrial toxins and mitochondria are known to be significantly injured in Reye's disease; therefore, it seems wise to avoid the use of aspirin in children during outbreaks of Reye's disease.