This study evaluated the efficacy of switching to didanosine in patients who were clinically or immunologically progressing despite zidovudine therapy. This multicenter, open-label study involved 400 patients with the acquired immunodeficiency syndrome (AIDS) or AIDS-related complex (ARC), who had tolerated zidovudine for at least 12 weeks and had signs of clinical or immunological disease progression. They were randomly assigned to receive 600 mg/d of zidovudine (n=133), 500 mg/d of didanosine (n=131), or 200 mg/d of didanosine (n=136). The primary end point was a new AIDS-defining event or death. The study was prematurely terminated, after the first interim analysis, mainly owing to results of two controlled studies demonstrating that a change to didanosine was associated with an improved outcome in patients with advanced HIV-1 disease. The median duration of follow-up was 53 weeks. The primary end point rates were 41, 58, and 59 (per 100 person-years) in the didanosine 500 mg, didanosine 200 mg, and zidovudine groups (zidovudine vs. didanosine 500 mg, relative risk 1.28, 95% confidence interval, 0.88-1.86, p = 0.19; didanosine 200 vs. 500 mg, relative risk 1.24, 95% confidence interval, 0.85-1.79, p = 0.26). In subjects with a baseline CD4 count of 100/mm3 or more, the primary end point rates were 8, 29, and 25 (per 100 person-years) in the didanosine 500 mg, didanosine 200 mg, and zidovudine groups, respectively (zidovudine vs. didanosine 500 mg, relative risk 2.96, 95% confidence interval 0.91-9.62, p = 0.07). No difference was seen in survival. In the didanosine 500 mg group, more patients had a 50% increase in CD4 cells (10% vs. 1% in zidovudine group, p = 0.01) and an increase of > or = 2.5 kg in body weight (2% versus 3%). Fatal pancreatitis developed in one patient assigned to didanosine 500 mg and in one to zidovudine. Our data suggest that switching from zidovudine to currently recommended doses of didanosine in subjects with ARC or AIDS who show evidence of clinical and laboratory disease progression can be associated with improvements in clinical outcome as well as in surrogate markers of HIV disease progression. This effect tended to be greater among individuals with higher CD4 counts (>100/mm3).