As part of a phase I/II trial in children infected with human immunodeficiency virus, we studied the pharmacokinetics of zidovudine and didanosine administered as single agents and in combination. Zidovudine (60 to 180 mg/m2 per dose) was given orally every 6 hours, and didanosine (60 to 180 mg/m2 per dose) every 12 hours. Pharmacokinetic samples were obtained from 54 patients and the area under the plasma concentration-time curve (AUC) was estimated by means of a previously defined limited sampling strategy. Follow-up blood samples were obtained after 4 and 12 weeks of treatment. The mean AUC for zidovudine ranged from 4.8 mumol.hr per liter at 60 mg/m2 to 11.0 mumol.hr per liter at the 180 mg/m2 level, and increased in proportion to the dose. The mean AUC for didanosine ranged from 2.8 mumol.hr per liter (60 mg/m2) to 8.0 mumol.hr per liter (180 mg/m2), with a wide interpatient variability. The AUCs of zidovudine and didanosine remained unchanged when the agents were administered in combination. There was no significant change in the AUCs of either drug after 4 and 12 weeks in comparison with those on day 3 of therapy. However, there was greater interpatient and intrapatient variability with didanosine than with zidovudine. These observations have implications for the future utility of therapeutic drug monitoring with these agents.