Field evidence suggests that agents belonging to the same group tend to behave similarly, i.e., behavior exhibits social interaction effects. Testing for such effects raises severe identification problems. We conduct an experiment that avoids these problems. The main design feature is that each subject simultaneously is a member of two randomly assigned and economically identical groups where only members ( neighbors') are different. In both groups subjects make contribution decisions to a public good. We speak of social interactions if the same subject at the same time makes group-specific contributions that depend on their respective neighbors' contribution. Our results are unambiguous evidence for social interactions. A majority of subjects is very strongly influenced by the contributions of their respective neighbors. Roughly ten percent exhibit no social interactions.