Lymphokine-activated killer cells (LAK) are cytolytic lymphocytes with the unique capacity of killing NK-resistant fresh human tumor cells in short-term assays. LAK appear to kill autologous tumors as well as TNP-modified self and allogeneic tumors with complete crossreactivity, both at the population and clonal level. Initial studies on the classification of LAK conclude that LAK are distinct from the classical NK and T-lymphocyte systems based on a number of criteria including surface phenotype, activation conditions, and spectrum of susceptible target cells. LAK kill rasoncogene-transfected fibroblasts in a manner similar to fresh tumors. As yet, the target cell determinant responsible for susceptibility to LAK lysis is unknown, but cell-surface proteins are definitely involved. Activation of LAK requires only IL-2, and is most efficient using serum-free conditions. Because interleukin-2 alone is sufficient for LAK activation, we have tested in vitro whether fresh PBL could be activated in the presence of tumor, as might be desired in vivo. LAK activation was greatly suppressed by tumor presence. LAK activation is also suppressed by hydrocortisone, but not cyclosporine A. Because of the above and other findings, we have initiated a clinical protocol to test whether LAK made from brain-tumor patients' PBL could eliminate residual glioma tumor cells. Autochthonous LAK, plus rIL-2 to maintain lytic ability, are injected during surgery. Preclinical studies in a rat glioma model have shown this approach to be safe. Eleven glioma patients have been injected intracerebrally with IL-2 and/or LAK with no immediate or long-term (14 months follow-up) adverse effects. Much work is needed to understand the LAK phenomenon and to resolve its potential usefulness in cancer therapy as well as its inherent biologic role.