To understand whether disease caused by the human immunodeficiency virus (HIV) affects zidovudine disposition, we compared the drug's pharmacokinetics in six healthy volunteers; six persons with the acquired immunodeficiency syndrome (AIDS) and no evidence of gastrointestinal (nausea, vomiting, diarrhea), renal (elevated blood urea nitrogen, serum creatinine), or hepatic (elevated liver function tests) disease; and three patients with AIDS and hepatic disease. After a single oral dose of zidovudine, serial blood samples were analyzed for drug concentration by radioimmunoassay. A one-compartment oral absorption model was fit to the concentration-time data. The absorption rate constant (4.05 vs 2.11 hr-1) and time to maximum concentration (0.61 vs 1.03 hr) were significantly different in healthy volunteers versus patients with AIDS without hepatic disease. Differences in half-life, oral clearance, and area under the curve were not statistically significant. In the three patients with AIDS plus hepatic disease, clearance was reduced an average of 63%, and area under the curve was increased by a factor of 2.3. These comparative pharmacokinetic data do not support profound differences between zidovudine's disposition in healthy volunteers and individuals with AIDS; however, the differences and trends that were observed may represent an effect of HIV disease. Although the presence of hepatic disease clearly indicates a need to modify individual dosages, these pharmacokinetic data may have more generalized implications for zidovudine dosing as the relationships between drug concentration and therapeutic or toxic effects are clarified.