This paper investigates bandwagon dynamics in social networks, using an extension of Granovetter’s (1978) threshold model. The focus is on the social ties connecting actors of different participation thresholds. A benchmark model is built on the principle of homophily, where actors are maximally similar to their network neighbors. Computational experiments show that bandwagon effects increase when network structures depart from pure homophily, allowing actors to have some new neighbors with more discrepant thresholds. Further increasing the heterogeneity of network neighbors causes the bandwagon effects to collapse, however, suggesting that a balance of heterogeneity and homogeneity of neighbors is optimal for bandwagon dynamics in social networks. This principle is compatible with insights from several empirical studies and consistent with the conclusions of other formal models.