Several potential mechanisms for viral destruction of HIV-infected cells have been described. The hypothesis was examined that if HIV were cytopathic, a positive relation between the in vivo virus production or CTL activity and infected cell death should be observed. In a regression analysis no significant relation was found between surrogate markers for in vivo virus production or the virus-specific CTL response and death rates of productively infected cells. In a subgroup of patients the hypothesis is rejected that HIV replication elicits a large (R(2) > 0.25) cytopathic effect (P < 0.05, N = 36). It is concluded that HIV replication elicits little cytopathic effect in productively infected cells and that CD4(+) T lymphocytes are eroded by other mechanisms.