Potentially innovative ideas are being generated, shared, and even remixed (recombined) in the online innovation communities. These ideas create new innovations through remixing of ideas. In this study, we investigate how remixing makes ideas more innovative in online innovation communities. Our model is validated through ordinary least squares regression on a secondary dataset of 57,049 ideas collected from one of the largest 3D printing online innovation communities, Thingiverse.com. The result shows that the number of prior ideas has an inverted U-shaped relationship with the idea's degree of innovation and the cross-boundary remix has a positive effect on the idea’s degree of innovation.