The precursors of murine lymphokine-activated killers (LAK) can be divided into two major subsets: NK-like (CD8-, NK1.1+, asialo GM1+) and T-like (CD8+, NK1.1-, asialo GM1+). LAK effectors have generally been characterized as being either CD8+ or NK1.1+. In this study, we divided each of these effector subsets further by virtue of their expression of B220 (as defined by the mAb 6B2) and Ly-24 (Pgp-1). Freshly obtained CD8+ and NK1.1+ cells were found, by fluorescence analysis, to be B220-. Lytically active LAK effector subsets were either CD8+ B220+ Ly-24+ or NK1.1+ B220+ Ly-24+. Most interestingly, a distinct NK1.1+ B220- Ly-24+ subset existed but had minimal lytic activity, suggesting that only a subset of NK cells is capable of acquiring the broad lytic activity of LAK. The acquisition of the B220 marker by the CD8+ subset closely paralleled its expression of lytic activity. However, classical MHC-restricted CD8+ CTL were Ly-24+ but remained B220- suggesting that the acquisition of the B220 marker, as defined by the 6B2 mAb, is not merely the result of cellular differentiation but may serve as a marker of MHC-nonrestricted killers. Three "classes" of target cells were examined for their susceptibility to lysis by the LAK effector subsets: YAC-1 (NK sensitive), CL27A (NK resistant), and autologous lymphoblasts that have been modified with 2, 4, 6-trinitrobenzene sulfonic acid. YAC-1 was lysed exclusively by the NK1.1+ B220+ Ly-24+ subset, 2,4,6-trinitrobenzene sulfonic acid-self was lysed exclusively by the CD8+ B220+ Ly-24+ subset whereas CL27A was lysed by both subsets. This pattern of lysis was confirmed by the in vivo depletion of NK1.1+ cells. It, therefore, appears that LAK effector subsets may be more selective in their lytic repertoire than previously thought.