Biological invasions pose one of the most severe environmental challenges of the twenty-first century. A longstanding idea is that invasion risk is predictable based on the phylogenetic distance – and hence ecological resemblance – between non-native and native species. However, current evidence is contradictory. To explain these mixed results, it has been proposed that the effect is scale-dependent, with invasion inhibited by phylogenetic similarity at small spatial scales but enhanced at larger scales. Analyzing invasion outcomes in a global sample of bird communities, we find no evidence to support this hypothesis. Instead, our results suggest that invaders are locally more successful in the presence of closely related and ecologically similar species, at least in human-altered environments where the majority of invasions have occurred. Functional trait analyses further confirm that the ecological niches of invaders are phylogenetically conserved, supporting the notion that successful invasion in the presence of close relatives is driven by shared adaptations to the types of niches available in novel environments.