Our objective was to determine whether a pharmacokinetic interaction exists between zidovudine and didanosine when they are coadministered. This was designed as a randomized, three-period, three-treatment, six-sequence, crossover study with a 1-week washout period between treatments. The patients were six men infected with human immunodeficiency virus who were asymptomatic. On three separate occasions, patients received zidovudine alone (200 mg every 8 h) for 3 days, didanosine alone (200 mg every 12 h) for 3 days, or zidovudine and didanosine for 3 days. On the fourth day, each patient received the final dose of each regimen, and blood and urine were serially collected for 8 h. Pharmacokinetic parameters were assessed for zidovudine, its glucuronide metabolite (GZDV), and didanosine. Coadministration of zidovudine had no significant effect on didanosine pharmacokinetic parameters (< 12% difference between treatment means, p > 0.1). Coadministration of didanosine did not significantly alter zidovudine pharmacokinetic parameters but did cause statistically significant increases in the renal and apparent formation clearances of GZDV (18.5% and 30.5% difference between the treatment means, respectively, p < 0.025). Therapeutic doses of zidovudine did not alter didanosine pharmacokinetic parameters. Coadministration of didanosine did not affect zidovudine parameters but did cause small alterations in GZDV pharmacokinetic values. These changes are unlikely to be clinically significant.