An experimental system has been developed to study in cloned lines of cells the control of Fc and C3 receptors by different compounds. The cells used were clones of mouse myeloid leukemic cells and the compounds used were the protein MGI2 (macrophage and granulocyte inducer) in serum from mice injected with bacterial endotoxin and the steroid inducer (SI) dexamethasone. Eight clones were isolated which could be divided into three groups. One group (MGI+SI+) was induced to form EA and EAC rosettes by MGI and only EAC rosettes by SI, the second group (MGI-SI+) was not inducible by MGI but was induced by SI to form EA or EA and EAC rosettes, and the third group (MGI-SI-) was not inducible for EA or EAC by MGI or SI. There were two types of MGI+SI+ clones, one type (D+) could be induced by MGI to differentiate to mature macrophages and granulocytes, and the other type (D-) could not be induced to differentiate to mature cells. The MGI-SI+ and MGI-SI- clones were all D-. The results indicate that there are different cellular sites for MGI and SI and that induction of EA and EAC rosettes did not seem to be mediated by cyclic AMP. Experiments with specifically bound 3H-BSA-anti-BSA complexes have indicated that there was an increase in the amount of 3H-BSA-anti-BSA bound per rosette-forming cell following induction by MGI or SI, and there were differences in the amount of 3H-BSA-anti-BSA bound per rosette-forming cell in different clones. These clones also showed differences in the shape of the curve for the number of EA rosette-forming cells obtained with erythrocytes coated with decreasing concentrations of antibody. The results suggest that such curves and those obtained with EAC rosettes can be used to determine the relative abundance of EA and EAC receptors on rosette-forming cells. EA rosettes on the myeloid leukemic cells, like those on normal macrophages and granulocytes, were specifically inhibited by IgG2a and by the Fc but not the Fab fragment of IgG. The EAC rosettes were inhibited by destroying the C3 component of complement. The different clones maintained their specific properties for at least 6 months in culture. The present system should, therefore, also be useful for studies on the genetic control of the regulation of Fc and C3 receptors.