Agarose is known to inhibit nonspecifically the cytotoxic effects of normal sera on xenogeneic lymphoid cells. To find an explanation for this agarose effect we have studied its requirements using guinea pig and human sera as the source of activity and rat thymocytes as target cells. Control assays were performed using heat-inactivated (56 C, 30 min) normal rat serum. The inhibitory effect of agarose was readily reproduced with untreated sera and also when sodium ethyleneglycoltetraacetate, a selective chelator of calcium ions, was added to the sera together with excess magnesium. However, the agarose effect failed to occur in the presence of 0.01 M EDTA unless magnesium ions were restored. Abrogation of cytotoxicity in human serum by incubation with a large number of target cells instead of agarose was also found to be magnesium dependent. Titrations of human serum, performed after absorption with agarose in the presence of EDTA, which does not interfere with antigen-antibody binding, and subsequent restoration of divalent cations, revealed no significant change in its cytotoxic titer when compared with that of mock-absorbed serum not subjected to the agarose treatment. Incubation of human serum with either agarose or rat thymocytes resulted in the conversion of factor B, essential for complement activation via the alternative pathway, previously shown to provide the complement activity necessary for the cytotoxic reaction. These results suggest that the agarose effect is mainly attributable to complement consumption via the alternative pathway rather than to the absorption of "natural" antibodies.