Maintaining ecological stability is essential for sustaining ecosystem functions and the benefits they provide to society. Ecological theory predicts that plant diversity either stabilizes or destabilizes local populations, while empirical studies report variable effects. We hypothesize that this discrepancy arises to a meaningful extent from differences in the ecological processes captured by various diversity and stability metrics. Analyzing over 8,000 permanent vegetation plots across biomes on five continents, we found a negative (i.e., destabilizing) diversity–stability relationship when using abundance-weighted rather than unweighted measures of population stability, which are more influenced by dominant species. Similarly, cumulative richness—capturing total species occurrence over time and long-term turnover—reveals a stronger destabilizing effect compared to average annual richness. Our findings reveal that, when specific metrics of diversity and stability are considered, increased interspecific coexistence tends to destabilize populations across natural ecosystems worldwide—particularly those of dominant species.