Workforce diversity is increasing across the globe, while organizations strive for equity and inclusion. Therefore, research has investigated how team diversity relates to performance. Despite clear arguments why diversity should enhance (some types of) performance, and promising findings in individual studies, meta-analyses have shown weak main effects. However, many meta-analyses have failed to distinguish situations where diversity should have a positive impact from those where its impact is more likely to be negative, leaving boundary conditions unclear. Here, we summarized the growing literature across disciplines, countries, and languages through a reproducible registered report meta-analysis on the relationship between diversity and team performance (615 reports, 2638 effect sizes). Overall, we found that the average linear relationships between demographic, job-related and cognitive diversity, and team performance are significant and positive, but insubstantial (|r|<.1). Considering a wide range of moderators, we found few instances when correlations were substantial. However, context matters. Correlations were more positive when tasks were higher in complexity or required creativity and innovation, and when teams were working in contexts lower in collectivism and power distance. Contrary to expectations, the link between diversity and performance was not substantially influenced by teams' longevity or interdependence. The main results appear robust to publication bias. Further research is needed on how diversity climates and team cultures affect these relationships, and when there may be non-linear relationships -- yet for the moment, promises of wide-spread performance increases may not be the strongest arguments to promote diversity initiatives. We discuss further implications for researchers and practitioners, and provide a web app to examine subsets of the data: https://lukaswallrich.shinyapps.io/diversity_meta/.