Social preference research has received considerable attention in recent years. Researchers have demonstrated that the presence of people with social preferences has important implications in many economic domains. However, it is important to be aware of the fact that the empirical basis of this literature relies to a large extent on experiments that do not provide anonymity between experimenter and subject. It has been argued that this lack of experimenter-subject anonymity may create selfish incentives to engage in seemingly other-regarding behavior. If this were the case these experiments would overestimate the importance of social preferences. Previous studies provide mixed results and methodological differences within and across studies make it difficult to isolate the impact of experimenter-subject anonymity on prosocial behavior. In this paper we use a novel procedure that allows us to examine the impact of the exact same ceteris-paribus variation in anonymity on behavior in three of the most commonly used games in the social preference literature. Our data does not support the hypothesis that introducing experimenter-subject anonymity affects observed prosocial behavior. We do not observe significant effects of experimenter-subject anonymity on prosocial behavior in any of our games.