OBJECTIVE To examine the effect of didanosine (2',3'-dideoxyinosine, ddI) on surrogate markers of HIV infection (CD4+ lymphocyte count, p24 antigen) and to evaluate the incidence of adverse effects from ddI. METHODS This study was performed as a retrospective chart review of patients who were enrolled in Bristol-Myers Squibb's expanded-access program for ddI. METHODS Patient records were obtained from primary care physicians' offices. METHODS Twenty-five HIV-infected patients diagnosed with AIDS or AIDS-related complex (ARC) who were intolerant of zidovudine (ZDV) therapy or deteriorating clinically despite ZDV therapy and were eligible for inclusion in the ddI expanded-access program. METHODS ddI was administered orally in a citrate-phosphate buffer every 12 hours. Patients were followed by their private physician on a monthly basis. METHODS Laboratory analysis at each month included CD4+ lymphocyte count, hemoglobin, hematocrit, serum amylase, uric acid, serum triglycerides, and p24 antigen. Mean CD4+ cell count, serum amylase, hemoglobin, and uric acid at each month during ddI therapy were compared with baseline concentrations for nine months. RESULTS Patients had received prior ZDV therapy for an average of 15.5 months before starting ddI. Mean CD4+ cell counts increased from 53.9/mm3 at baseline to 72.4/mm3 after 4 months of therapy (p = 0.04) but returned to concentrations comparable with those at baseline after 5 months. One case of documented pancreatitis, two cases of suspected pancreatitis, and nine cases of peripheral neuropathy occurred during ddI therapy. The estimated mean cumulative dose for the development of neuropathy was 1.16 g/kg, which is lower than previously reported. CONCLUSIONS Patients who have received prolonged therapy with ZDV or who have low initial CD4+ counts may not have sustained increases in CD4+ counts from ddI therapy. Also, development of peripheral neuropathy may occur at lower cumulative doses in these patient populations.