Crowdsourcing is an effective and powerful tool for firms to solve practical issues and develop innovative products, but the relationship between risks and performance in crowdsourcing has received insufficient attention. Based on a dataset of 163 samples from China, social subsystem risk is empirically found to negatively influence crowdsourcing performance, whereas technical subsystem risk affects the performance insignificantly. The negative impact of social subsystem risk on performance is stronger than that of technical subsystem risk. These findings reveal that different types of risk have diverse roles in affecting performance. Moreover, we provide novel knowledge to existing literature by empirically indicating that different risk types interact with one another to influence performance. Technical subsystem risk particularly enhances the negative effect of social subsystem risk on performance. Therefore, crowdsourcers should develop systematic but different risk management strategies to mitigate the two risk types.