The kinetics of activation, cell surface phenotype, target cell specificity and anatomic localization of pulmonary natural killer cells was examined in (C57BL/6 X A/J)F1-hybrid mice with acute graft-versus-host reactions induced by intravenous injection of 50 x 10(6) parental strain lymph node and spleen cells. Results showed that there was a marked increase in NK cell activity directed against YAC-1 tumor cells. This activity remained elevated in the lung over almost the entire course of the reaction, whereas it was only transiently increased in the spleen during the early stages of the reaction and then fell to control values. During the reaction, NK cells from both organs acquired the ability to kill P815 targets, cells that are normally insensitive to NK cell lysis. The level of P815 killing never reached that achieved against YAC-1 cells, but was significantly higher in the lung than in the spleen. Antibody and complement depletion experiments showed that both anti-YAC-1 and anti-P815 activity could be depleted with antiserum to the asialo-GM1 cell surface marker, but was unaffected by anti-Lyt-1.2 and anti-Lyt-2.2 treatment. Anti-YAC-1 activity was partly sensitive to depletion with anti-Thy-1.2. Cytotoxicity to P815 target cells acquired during the reaction was completely abrogated by anti-Thy-1.2. These findings suggest that during the reaction two phenotypically distinct types of NK cells are activated: a conventional, Thy-1-negative cell that kills only YAC-1 targets, and a Thy-1-positive cell with a broadened spectrum of lytic activity. We suggest that the latter may be generated in response to interleukin-2 released during the lymphoproliferative phase of the reaction and may represent a type of lymphokine-activated killer cell. Our results revealed that virtually all of the NK cell activity in the lung could be attributed to cells residing in the interstitium, or to cells tightly adherent to endothelium or epithelium. There was no correlation between augmented NK cell activity in the lung and the presence of peribronchial and perivascular mononuclear cell infiltrates seen in histological sections of the lung. These findings do not appear to support the idea that NK cells are by themselves directly responsible for the pathological changes produced by the reaction.