The objective of this study was to compare the effects of zidovudine and didanosine on health-related quality of life in persons with advanced HIV infection and varying duration of prior zidovudine exposure. It was designed as a substudy nested in two similar placebo-controlled active-control-arm randomized trials, using sites of the AIDS Clinical Trials Group participating in the randomized trials of zidovudine versus didanosine (ACTG 116 and 117). The patients comprised 356 participants enrolled in ACTG 116 and 117. All had HIV infection and either a CD4 count of <200 cells/mm3, or a CD4 count of <300 cells/mm3 plus symptoms of HIV disease. Participants were randomized equally within strata defined by duration of prior zidovudine therapy, to receive didanosine sachets at a dose of 500 mg daily (334 mg in subjects weighing <60 kg) or 750 mg daily (500 mg in subjects weighting <60 kg) plus inactive capsules resembling zidovudine, or to receive zidovudine capsules at a dose of 600 mg daily plus inactive sachets resembling didanosine. The main outcome measures were self-reported health-related quality of life, healthcare utilization, disability, work and symptom impact. The results showed no differences in reported symptom impact or healthcare utilization, and most measures of disability were similar. In the group with more than 8 weeks of prior zidovudine therapy, several of the health status scale scores for ongoing participants were significantly better for didanosine recipients, but average differences were small. Use of several different approaches to combining health status and survival showed no differences in the overall quality-time experiences between the treatment groups. Individuals taking zidovudine, low-dose didanosine and high-dose didanosine experienced 33, 34 and 35 weeks, respectively, in at least the typical health state if they had fewer than 8 weeks of previous zidovudine therapy, and had 23, 23 and 26 weeks, respectively, if they had more than 8 weeks previous use of zidovudine. Results did not differ when data were analysed within strata ofpatients who had any versus no prior exposure to zidovudine, or AIDS versus non-AIDS status. In conclusion, functional status and health-related quality of life were substantially similar among persons receiving either zidovudine or didanosine, regardless of the duration of prior zidovudine treatment.