NK cell activity, defined by the ability of infiltrating host cells to lyse the YAC-1 tumor target, can be detected in sponge matrix allografts across all genetic barriers tested. Nonspecific tumor cell killing cannot be detected either within bulk populations of host-infiltrating cells or in populations enriched for non-adherent lymphocytes. NK activity is also detected in cells infiltrating a syngeneic sponge matrix graft although to a much lesser extent than an allogeneic graft. NK cell functional activity at the graft site precedes the appearance of alloimmune CTL by several days. The surface phenotype of the NK cell is Thy-1.2+ and L3T4- as determined by depleting the various subpopulations with antibody and C. Systemic treatment of sponge-bearing animals with repeated injections of anti-asialo GM1 (AGM1) results in inhibition of both NK activity and CTL activity recovered from the graft on days 5 to 9 after grafting, but on days 11 to 13 after grafting both NK activity and CTL activity are found within the sponge graft. Treatment of sponge-associated cells with anti-AGM1 in vitro or intrasponge injection of anti-AGM1 at various times after grafting eliminates NK activity more readily than alloimmune CTL activity. The intimate association observed between NK cells and alloimmune CTL at the graft site prompts further investigation into the role of NK cells in the allograft response.