Contests are frequently used to raise the workers’ productivity and innovation in business, government, and many other settings. They can take many different formats or designs, but two seem prevalent: the race and the tournament. Races set the incentives by rewarding the first person to meet a specified, typically hard-to-reach, technical goal. Tournaments set the incentives by rewarding the competitor attaining the highest performance in a given period. In this study, we use a field experiment on a popular crowdsourcing platform (Topcoder) to measure empirically the differences in performance and participation of competitors randomly assigned to either a race or a tournament competition.