Thymopentin, 50 mg subcutaneously (s.c.) 3 times per week, was evaluated in a double-blind, randomized, placebo-controlled trial of zidovudine (AZT)-treated asymptomatic human immunodeficiency virus (HIV)-infected subjects with 200-500 CD4 cells/mm3 at entry. The 352 subjects were prestratified by prior AZT use into stratum I (235 subjects, > 6 months AZT at entry) and stratum II (117 subjects, < or = 6 months AZT at entry). Clinical end points, CD4 cell counts, serum p24, serum immune complex dissociated (ICD) p24, and safety variables were evaluated through 48 weeks, using an intent-to-treat analysis. The two strata were analyzed individually because they yielded different clinical outcomes, with a statistically significant treatment-by-stratum interaction. In stratum I (mean, 16 months AZT at entry) two AIDS or death events occurred in thymopentin and 10 in placebo recipients (p = 0.024; relative risk (RR) estimate, 4.9 [95% confidence limit (CI), 1.1 to 22.2]). There were three AIDS-related complex (ARC), AIDS, or death events in thymopentin and 18 in placebo recipients [p = 0.001; RR estimate, 5.9 (95% CI, 1.7 to 20.0)]. In stratum II (mean, 3 months AZT at entry), four AIDS or death events occurred in thymopentin and none in placebo recipients (p = 0.11), and four ARC, AIDS, or death events occurred in thymopentin and two in placebo recipients (p = 0.79). The treatment groups did not differ significantly with respect to changes in CD4 counts or p24 antigen levels or with respect to clinical adverse experiences or laboratory abnormalities. Thus, AZT-experienced placebo-treated subjects had relatively high progression rates to AIDS or death and to ARC, AIDS, or death, and these rates were reduced by thymopentin treatment. In contrast, placebo-treated subjects with little prior AZT experience had low progression rates; these were not significantly changed by thymopentin treatment. There was no increase in the incidence of adverse reactions with thymopentin.