A 64-year-old white man, who had never received a transfusion, was found to have anti-M in his serum. The antibody agglutinated all M+ red cells in room-temperature tests. When the ionic strength of the test milieu was reduced by use of an additive solution and the tests were incubated at 37 degrees C, the antibody hemolyzed M + N- but not M+N+ red cells. All M+ red cells reacted in indirect antiglobulin tests using polyspecific antiglobulin reagents when such tests followed an initial incubation at room temperature. When red cells and the patient's serum were warmed to 37 degrees C before being mixed, no antibody activity was demonstrable. The antibody was adsorbed to exhaustion onto M+N- and M+N+ red cells (including the patient's own), and its activity was destroyed by dithiothreitol. There was no evidence of in vivo red cell destruction by the autoantibody. No previously reported example of anti-M has been shown to activate complement in conventional in vitro tests. This example was extraordinary in that it caused sufficient complement activation to present as an in vitro hemolysin.