Bacteria whose lipopolysaccharide contains O-antigen side chains activate complement via the alternative pathway. We have shown previously that three strains of Salmonella, differing in the chemical structure of their O-antigens, consumed C3 to different extents when incubated in C4-deficient guinea pig serum. Moreover, sheep erythrocytes coated with lipopolysaccharide purified from these strains mimicked whole cells in C3 consumption, proving that lipopolysaccharide alone could account for these results. We have now measured the deposition of 125I-C3 in this system, and found that C3 deposition parallels C3 consumption in rate and extent, and differs for surfaces bearing different O-antigens, whether tested with bacteria or with erythrocytes coated with purified lipopolysaccharide. We have also examined the fate of C3 on these Salmonellae by measuring the size and quantity of 125I-C3 breakdown fragments by SDS-PAGE, and have determined the kinetics of conversion of C3b to iC3b by using conglutinin, a molecule that binds specifically to iC3b. There is no difference in breakdown of C3b deposited on cells with different O-antigens: all show partial conversion to iC3b and C3dg as indicated by 68,000, 44,000, and 41,000 m.w. bands on reduced SDS gels. Furthermore, for all strains, the Ka of conglutinin binding to iC3b is similar (0.49 to 0.69 X 10(8) M-1), as is the rate of generation of iC3b and the final ratio of iC3b:C3b + iC3b (0.62 to 0.72). We therefore postulate that the fine structure of the O-antigen in lipopolysaccharide determines the magnitude of alternative pathway activation on the bacterial surface by affecting the rate and extent of C3b deposition, but not the rate and extent of breakdown of C3b.