To assess the efficacy and safety of thymopentin in HIV-infected patients who had not yet developed AIDS. Patients were stratified into asymptomatic or symptomatic groups and randomized to receive either thymopentin (50 mg) or placebo, subcutaneously, double-blind for 24 or 52 weeks, three times a week. Patients were enrolled at three sites (two hospital clinics and one private practice). Of 91 HIV-seropositive patients (52 asymptomatic and 39 symptomatic) from whom HIV could be isolated from peripheral blood, 45 were enrolled for 24 weeks and 46 for 52 weeks of double-blind evaluation. Virological, immunological and clinical evaluations were performed before and during treatment. Thymopentin-treated asymptomatic patients had more CD4+ cells, as demonstrated by a greater area under the percentage CD4+ cells curve (P = 0.03) and a shorter median time to a 20% increase in percentage of CD4+ cells (P = 0.04) in the first 24 weeks, with similar trends in the 52-week study. By 24 weeks no asymptomatic thymopentin-treated and two placebo-treated patients (9.1%, Kaplan-Meier estimate) had progressed to constitutional symptoms (P = 0.12; two-tailed Wilcoxon-Gehan test), with only one further progression in a placebo-treated patient in the subset followed for 52 weeks. Symptomatic patients receiving thymopentin or placebo were similar in both CD4+ cell levels and disease progression (two progressions to AIDS in each group). No serious adverse effects attributable to thymopentin were observed. These results, if confirmed, indicate that thymopentin, by maintaining CD4+ cells, could slow or arrest immune decline and consequent disease progression at the asymptomatic stage of HIV infection.