2',3'-Dideoxyinosine (ddI) has shown activity against human immunodeficiency virus in phase I clinical trials. The drug is rapidly degraded by acid, however, thus raising questions as to the efficiency and reproducibility of its absorption after oral administration. This investigation studies the bioavailability of several oral dosage forms of ddI. When ddI was given to fasting patients as an oral solution with antacid, the bioavailability was 41% +/- 7% (mean +/- SEM). However, when given as buffered tablets, the bioavailability was considerably less (25% +/- 5%). The bioavailability increased slightly when the tablets were given with supplemental antacid (36% +/- 6%). Two enteric-coated preparations had reasonable bioavailability (36% +/- 5% and 26% +/- 5%), but the peak plasma level was much lower and occurred at a much later time than with the oral solution. When ddI was given as a premeasured powder containing sucrose and buffer to be reconstituted by the patient (the "sachet" preparation), the bioavailability was 29% +/- 6%. This was similar to that of the oral solution for this particular group of patients (30% +/- 7%). However, the bioavailability of the sachet was only 17% +/- 4% when administered with food. When the sachet was given to patients receiving ranitidine, no consistent change in bioavailability was noted. Also, no change in ddI pharmacokinetics was noted in patients receiving ganciclovir.