It was previously demonstrated that preimmunization of mice with live vaccinia virus (VV) and subsequent immunization with VV-infected (modified), syngeneic tumor cells resulted in enhanced induction of tumor-specific immunity through a cellular collaboration between VV-reactive helper T (VV-Th) cells and tumor-specific effector cell precursors. On the basis of this augmenting system, a tumor-specific immunotherapy model was established in which a growing tumor regressed. C3H/HeN mice were pretreated with cyclophosphamide to eliminate nonspecific suppressor T cell activity and subsequently inoculated (primed) s.c. with live VV, leading to augmented induction of VV-Th cell activities. Four weeks later, the mice were inoculated i.d. with syngeneic X5563 myeloma cells. Six days after the tumor cell inoculation, live VV was injected into the tumor mass three times at 2-day intervals. Seven of ten mice which had received VV priming and subsequent VV injection into the tumor mass exhibited complete tumor regression. On the contrary, mice which had received mere intratumoral VV injection in the absence of VV priming failed to exhibit appreciable tumor regression. Mice whose tumor had completely regressed (regressor mice) by the above VV immunotherapy were shown to have acquired systemic anti-tumor immunity, which was confirmed by a challenge with syngeneic tumor cells after the tumor regression. In vitro analysis of these immune mice revealed that potent tumor-specific cytotoxic T lymphocyte responses were preferentially induced, but with no detectable anti-tumor antibody responses. Such a potent tumor-specific immunity was not observed in mice which had received an intratumoral VV injection in the absence of VV priming. Thus the results clearly indicate that the tumor regression was accompanied by concurrent generation of a potent tumor-specific immunity, suggesting that the cellular collaboration between VV-Th cells and tumor-specific effector cell precursors is also functioning in this VV-immunotherapy protocol, likewise the immunoprophylactic model. Therefore, the present model provides an effective maneuver for tumor-specific immunotherapy and this system will be theoretically applicable to human cancer treatment.