Online innovation contest is a new effective way for open innovation. The sponsor, either corporate or individual, can submit contest problems online to seek solutions from a variety of solvers, and the solver who submits the best solution will get the award set by the sponsor. This new electronic commerce model is gaining more and more attention from researchers and practitioners. One of the serious problems in online innovation contest is sponsor s fraud behavior by registering two accounts in the same online contest platform and choosing himself as the winner while adopting solutions submitted by other solvers. Trust for sponsor which is critical factor influencing users behavior is reduced by sponsors fraud. This paper aimed to study: i) the antecedents of solvers trust in sponsor; ii) how contest platform set the optimal fraud detection effort. For the first issue, a research model based on trust theory is constructed, and a survey was undertaken to test the model. The research results indicate that the current reputation mechanism to foster trust does not influence solvers trust significantly, and the online platform’s credibility and the solvers satisfaction of past experience are vital to trust. The results from survey highlighted the importance of the online innovation contest platform service quality. For question two, a game theory model is constructed, and the optimal fraud detection effort balancing the reputation cost and detection cost is obtained with backward induction method. Implications and research extensions are identified to guide further research in this area.