Murine IL-2-induced lymphokine-activated killers (LAK) can be divided into two mutually exclusive subsets: NK1.1+CD8- and NK1.1-CD8+. We have previously established that the lytically active subset of each of these two populations expressed B220, as determined by the mAb 6B2, on its surface. In this study, we examined the lytically active subsets induced by IL-4. We found that, similar to IL-2, the lytically active IL-4-LAK expressed B220 on their surface. Similar to IL-2-LAK, IL-4-LAK cultures also contained NK1.1+CD8- B220+, and NK1.1-CD8+B220+ subsets. IL-4-LAK cultures, however, contained two novel and unexpected lymphocyte subsets determined by their surface markers to be either: i) NK1.1+CD8 alpha+beta-B220+, or ii) NK1.1-CD8 alpha+beta-B220+. These subsets were not derived from NK1.1+ or CD8+ precursors but were induced from a CD4-CD8-NK1.1-Slg- splenocyte subpopulation. Most of the CD8 alpha+beta- cells seen in the IL-4 cultures expressed TCR-alpha beta with little or no TCR-gamma delta detected on their surface. Similar to IL-2-LAK, the IL-4-LAK subsets appeared to display a bias as to their susceptible target cells with the NK1.1-CD8 alpha+beta+ subset being most potent against trinitrophenyl-modified autologous lymphoblasts (2,4,6-trinitrobenzene sulfonic acid (TNBS)-self). The NK1.1+CD8 alpha-beta- effectors were most potent against YAC-1 and CL27A with little activity against TNBS-self. All three targets were susceptible to lysis by the NK1.1+CD8 alpha+beta- subset. These findings document the characteristics of a novel lymphocyte subset, derived from splenic precursors, which shares certain features with cells previously thought to be present exclusively in intraepithelial lymphocytes.