<p>Although numerous studies have demonstrated a positive relationship between biodiversity and ecosystem stability, that is low temporal variability in ecosystem functions, the role of interspecific interactions in driving this relationship remains elusive. Here we develop a partitioning framework to disentangle the effects of interspecific interactions on ecosystem variability from those of interaction-independent processes. Applying this framework to competition models and biodiversity experiments in European and North American grasslands, we find that species-specific dynamics observable in monocultures (for example, environmental responses or demographic stochasticity), rather than interspecific interactions observable only in mixtures, explain much of the stabilizing effects of biodiversity on ecosystems. The weak net effect of interspecific interactions results from two counter-balancing effects: an interaction-driven increase in average species variability and an interaction-driven decrease in average species synchrony. We also find that interspecific interactions contributing to higher ecosystem functioning tend to reduce ecosystem variability. Our study provides a systematic empirical assessment of the role of interspecific interactions in shaping grassland ecosystem stability, challenging traditional assumptions about their importance and highlighting species-specific dynamics as the primary mechanism underlying the pervasive biodiversity–stability relationship.</p>