Studies on the effect of individual diversity on organization performance have gained much attention. However, it is not clear whether the insights derived from those studies can be directly applied in the settings of online innovation contests. Based on social categorization perspective and information/decision-making perspective aggregated from former literature, and behavioral data of 8 296 online innovation contests, it proposed and verified the inverted U-shape relationship between solvers’ country diversity and skill profile diversity and the contest performance. Additionally, following the positive or negative effects of diversity on contest performance, it proposed and tested that both “parallel path ” and “aiding learning” partly mediate the effect of diversity on the contest performance. The focal research contributes to a more comprehensive and deep understanding about the effect of diversity, and extends the research area from traditional organizations to online innovation contests. The significant effects in the focal research provide organizers of contest platforms or seekers of innovation contests a key managerial insight: there is an optimal level of solver diversity for contest performance, that is, too much is as bad as too little.