Natural killer (NK) activity in the rat and human has been attributed to cells having the morphology of large granular lymphocytes (LGL). However, this association has been less clear in the mouse, largely because of difficulties in obtaining highly enriched populations of LGL from normal spleen and blood. We have previously observed that the administration of the biological response modifier (BRM) maleic anhydride divinyl ether (MVE-2) strongly augmented NK activity in lung and liver, and the augmented NK activity coincided with increased resistance to the formation of experimental metastases in these organs. The degree of NK augmentation was most striking in the liver, an unexpected and previously unreported observation. In the present study, both MVE-2 or Corynebacterium parvum induced a dramatic augmentation of liver NK activity, which reached maximum levels 3-5 d after treatment. This augmentation of NK activity in the liver coincided with a large increase in the number of lymphoid cells with the morphological characteristics of LGL that could be isolated from enzymatically digested suspensions of perfused liver. The yield of LGL per liver following BRM treatment corresponded to a 10-50-fold increase as compared to normal mice. LGL were purified from these enzymatically digested suspensions of perfused liver by depletion of adherent cells on nylon wool columns and subsequent enrichment for low-density lymphoid cells by fractionation on Percoll density gradients. The enrichment of LGL correlated with greatly increased NK activity against YAC-1. Conversely, the higher-density fractions were depleted of both LGL and NK activity. This increase in NK activity in the liver was suppressed by in vivo treatment with anti-asialo GM1 (asGM1) serum. This treatment also resulted in a corresponding reduction in both the total number and percentage of LGL. By flow cytometry analysis, the phenotype of the majority of these highly cytolytic LGL isolated from the livers of BRM-treated mice were asGM1+, Thy-1+, Ly-5+, Qa-5+, Mac-1+, and Gma-1+, whereas these LGL were Ly-1-, Lyt-2-, L3T4-, and surface Ig-. We conclude that the livers of BRM-treated mice can provide a rich source of highly active mouse LGL that could be used for further characterization of this lymphocyte subset. Further, these studies imply a potential for BRM therapy of neoplastic or viral diseases through augmentation of organ-associated immune responses.