The purpose of this study is to analyze the impact of homophily on diffusion over social networks. An agent-based simulation model is developed to serve as the experimental ground for this analysis. Diffusion dynamics of a nonsticky innovation is investigated by varying homophily levels in the social network depicted in the model as the primary control variable. First of all, the results show that homophily is self-reinforcing. Second, starting from a nonhomophilous network, early increases in the level of homophily have a positive effect on the extent of diffusion, whereas further increases have a negative impact. Finally, several local minima and maxima are observed in the relation between the homophily level and the extent of diffusion. Our analysis focuses on node properties such as connectedness and average degrees in order to explain the observed regular relationship between homophily and diffusion. We argue that (i) homophily increases the connectedness of different status groups separately and (ii) increasing levels of homophily decreases the marginal importance of a single homophilous tie by increasing the sources of valuable information. Future research involves investigating the coevolution of social behavior and networks by allowing the adopted innovation to lead to value homophily, exploration of different diffusion initiation types, and different adoption heuristics.