This paper uses the random assignment of playing partners in professional golf tournaments to test for peer effects in the workplace. We find no evidence that the ability of playing partners affects the performance of professional golfers, contrary to recent evidence on peer effects in the workplace from laboratory experiments, grocery scanners, and soft-fruit pickers. In our preferred specification, we can rule out peer effects larger than 0.045 strokes for a one stroke increase in playing partners' ability, and the point estimates are small and actually negative. We offer several explanations for our contrasting findings: that workers seek to avoid responding to social incentives when financial incentives are strong; that there is heterogeneity in how susceptible individuals are to social effects and that those who are able to avoid them are more likely to advance to elite professional labor markets; and that workers learn with professional experience not to be affected by social forces. We view our results as complementary to the existing studies of peer effects in the workplace and as a first step towards explaining how these social effects vary across labor markets, across individuals and with changes in the form of incentives faced. In addition to the empirical results on peer effects in the workplace, we also point out that many typical peer effects regressions are biased because individuals cannot be their own peers, and suggest a simple correction.