Inflammatory action of the potent chemotaxin C5a has been well characterized on a variety of human cell types, including neutrophils, monocytes, basophils, and eosinophils. The cellular effects of C3a are less well defined. Contradictory reports have been published for C3a activation of neutrophils. Recent reports that C3a activates both basophils and eosinophils prompted us to reinvestigate the effects of C3a stimulation on eosinophils. We hypothesized that C3a activation of eosinophils, cells that are present in most neutrophil preparations, might lead to neutrophil activation. Using neutrophils of 98% purity, we observed no evidence of cellular activation after stimulation with either C3a, recombinant human C3a (rhC3a), or the synthetic C3a analogue C3a 57-77, Y57. Eosinophils purified to > 98% purity displayed concentration-dependent polarization, chemotaxis, and enzyme release by stimulation with C3a, rhC3a, and the synthetic C3a analogue. An inactive form of C3a, C3adesArg, failed to stimulate either eosinophils or neutrophils. Using neutrophil preparations containing 5-9% eosinophils, up to 20% of neutrophils became polarized after exposure to C3a. Likewise, we demonstrated that supernatant from C3a-stimulated eosinophils promotes neutrophil chemotaxis. Eosinophil polarization experiments were repeated in the presence of antibody to the C5a receptor (C5aR) to show that C3a and C5a interact with different receptors. C3a activates eosinophils in the presence of anti-C5aR antibody at concentrations that fully block C5a activation. We conclude that eosinophils are directly activated by either C3a or C5a, whereas C3a failed to activate neutrophils. C3a acts on eosinophils via a receptor that is distinct from C5aR. Since neutrophils are indirectly stimulated by C3a, eosinophils contaminating neutrophil preparations may explain earlier reports that C3a activates human neutrophils.