Lymphokine-activated killer (LAK) cells were generated from peripheral blood lymphocytes (PBL) that were depleted of mature cytotoxic natural killer (NK) cells. PBL NK activity was abolished by pretreatment of effector cells with the toxic lysosomotropic agent L-leucine methyl ester (LME) or by depletion of effector cells by K562 monolayer absorption (MA). Both treatments markedly reduced the proportion of cells expressing NK-associated markers such as CD 16 (Leu 11b, B73.1), Leu 7, and NKH-1 (Leu 19), whereas these treatments had minimal effects on cells expressing T cell markers (CD 3, CD 4, and CD 8). LME and MA also drastically decreased the proportion of K562 target-binding lymphocytes. LAK activity against NK-sensitive and NK-resistant targets can be generated from the NK cell-depleted PBL by incubation with interleukin-2. Peak LAK activity generated from MA-treated PBL was later than the peak of LAK activity generated from either untreated or LME-treated PBL. Although MA of PBL on NK-resistant S4 sarcoma targets had little effect on NK activity, LAK activity against both K562 and S4 targets was reduced. These results suggest that there are at least three LAK precursor subpopulations in PBL: mature NK cells that can bind and kill K562 targets (LME-sensitive and MA-sensitive); "pre-NK" cells that can bind but cannot kill (LME-resistant and MA-sensitive); and non-NK cells that cannot bind and cannot kill K562 targets (MA-resistant).