Tournament incentives prevail in labor markets, in particular with respect to promotions. Yet, it is often unclear to competitors how many winners there will be or how many applicants compete in the tournament. While it is hard to measure how this uncertainty affects work performance and willingness to compete in the field, it can be studied in a controlled lab experiment. We present a novel experiment where subjects can compete against each other, but where the number of winners is either uncertain (i.e., unknown numbers of winners, but known probabilities) or ambiguous (unknown probabilities for different numbers of winners). We compare these two conditions with a control treatment with a known number of winners. We find that ambiguity induces a significant increase in performance of men, while we observe no change for women. Both men and women increase their willingness to enter competition with uncertainty and ambiguity, but men react slightly more than women. Overall, both effects contribute to men winning the tournament significantly more often than women under uncertainty and ambiguity. Hence, previous experiments on gender differences in competition may have measured a lower bound of differences between men and women.