This article carries out a meta-analysis on empirical studies addressing cluster–innovation relationships since the 1980s. The results show that (1) clusters have positive effects on innovation; (2) different studies demonstrate heterogeneity in the estimated effect sizes; (3) several moderators are shaping the cluster–innovation relationship, for example, in what industries the cluster specializes, in which geographical region the cluster is located, and so on. This article reveals the cluster–innovation relationships are state-contingent and provides guidance on evaluating whether a cluster strategy can encourage innovation in a specific region. For example, the regression result indicates initiating a manufacturing cluster in a US region is expected to bring about fifteen more patents every year.