We exploit a natural experiment in which two professionals compete in a one-stage contest without strategic motives and where one contestant has a clear exogenous psychological momentum advantage over the other in order to estimate the causal effect of psychological momentum on performance. We find that men's performance is significantly affected by psychological momentum, while women's is not. This result is robust to different specifications and estimation strategies. Our results are in line with evidence in the biological literature that testosterone, which is known to enhance performance of both men and women, commonly increases following victory and decreases following loss only among men. Implications of our findings for contest design are also discussed.