Natural killer cell activity was evaluated in children with acute lymphocytic and acute myelogenous leukemia. Peripheral blood mononuclear cells isolated at the time of diagnosis and before initiation of therapy were mixed with 51Cr-labeled K562 or MOLT-4 target cells at a ratio of 100:1. In 13 consecutive cases of acute lymphocytic leukemia, the mean percentage of lysis of K562 cells (15.0%) was significantly below that of adult (49.8%) and age-related controls (35.9%). A similar pattern was observed against MOLT-4 targets (acute lymphocytic leukemia, 11.3%; adults, 39.8%; and pediatric controls, 28.4%). The mean activity in 8 cases of acute myelogenous leukemia was also markedly reduced (6.8% versus K562 and 6.0% versus MOLT-4). Linear regression analyses of white blood cell, lymphocyte, and leukemia blast counts failed to demonstrate any correlation between peripheral cell counts and natural killer cell activity. Thus, it would not appear that the observed decrease in lysis was due merely to dilution of effectors with blasts. The lytic activity of cells isolated from patient blood was significantly lower than that from cells isolated from an equal volume of blood from a normal adult. These results suggest that the decreased natural killer cell activity is not explained by simple dilution. Instead, they indicate an absolute decrease in lytic potential. Additional experiments have precluded suppressor cell involvement and competitive inhibition of blasts with target cells as possible causes for depressed lysis.