Declining CD4+ T-cell numbers and anti-CD3-induced T-cell responsiveness are prognostic markers for progression of HIV infection. We investigated the effect of long-term (2-year) zidovudine treatment on these immunological markers in a group of nine asymptomatic p24-antigenaemic men, five of whom progressed to AIDS. A group of 10 untreated HIV-infected men, five of whom progressed to AIDS, was studied as a control. At intake, 1 year before the start of treatment, CD4+ T-cell numbers in the groups were not significantly different. However, at that time progressors already exhibited an extremely low anti-CD3-induced T-cell responsiveness compared with non-progressors. In all people T-cell responsiveness and the number of CD4+ T-cells had improved 6 months after the start of zidovudine treatment. However, CD4+ T-cell numbers were not persistently elevated, and restoration of T-cell responsiveness was of only short duration. Our results show that zidovudine treatment in the asymptomatic phase of HIV infection did not result in a sustained improvement in T-cell function. Furthermore, they suggest that differences in clinical course among zidovudine-treated asymptomatics may be caused by heterogeneity of this group with respect to T-cell functional capacity at the start of treatment.