This report describes the one-year results of a noncomparative study designed to assess the safety and tolerance of low-dose zidovudine (azidothymidine) given orally to 60 human immunodeficiency virus type 1-infected infants and children. At baseline, the mean age was 1.9 years (+/- 1.4), and all were symptomatic: 43% were P2A and 57% were P2B to F according to the Centers for Disease Control classification. All the patients received zidovudine for at least 6 months, and 52 of them (87%) completed a full year of therapy. The mean duration of follow-up was 346 days (+/- 42) (range, 183 to 366 days). The initial therapy consisted of four daily doses of 100 mg/m2 (400 mg/m2 per day, equivalent to 20 mg/kg per day). However, this treatment was modified when neutropenia or anemia was observed. Twenty-nine children (48%) remained at the initial therapy for the entire study. Zidovudine dosage was adjusted 92 times in the other 31 children (52%), mostly due to neutropenia (83%). Altogether, the time under full-dose therapy represented 81% of the total duration of the protocol for all patients. Children with mild symptoms, P2A at study entry, were more likely to remain under full-dose therapy than children with severe symptoms, P2B to F: the time under full-dose therapy represented 91% of the duration of the protocol for the former group and only 74% for the latter one (P less than .02). No clinical adverse experiences were attributed directly to zidovudine. Thirty-seven children were prescribed trimethoprim-sulfametoxazole as a prophylaxis for Pneumocystis carinii pneumonia.(ABSTRACT TRUNCATED AT 250 WORDS)