The concept of ecological resilience (the amount of disturbance a system can absorb before collapsing and reorganizing) holds potential for predicting community change and collapse—increasingly common issues in the Anthropocene. Yet neither the predictions nor metrics of resilience have received rigorous testing. The cross-scale resilience model, a leading operationalization of resilience, proposes resilience can be quantified by the combination of diversity and redundancy of functions performed by species operating at different scales. Here, we use 48 years of sub-continental avian community data aggregated at multiple spatial scales to calculate resilience metrics derived from the cross-scale resilience model (i.e., cross-scale diversity, cross-scale redundancy, within-scale redundancy, and number of body mass aggregations) and test core predictions inherent to community persistence and change. Specifically, we ask how cross-scale resilience metrics relate community stability and collapse. We found low mean cross-correlation between species richness and cross-scale resilience metrics. Resilience metrics constrained the magnitude of community fluctuations over time (mean species turnover), but resilience metrics but did not influence variability of community fluctuations (variance in turnover). We show shifts in resilience metrics closely predict community collapse: shifts in cross-scale redundancy preceded abrupt changes in community composition, and shifts in cross-scale diversity synchronized with abrupt changes in community composition. However, we found resilience metrics only weakly relate to maintenance of particular species assemblages over time. Our results distinguish ecological resilience from ecological stability and allied concepts such as elasticity and resistance: we show communities may fluctuate widely yet still be resilient. Our findings also differentiate the roles of functional redundancy and diversity as metrics of resilience and reemphasize the importance of considering resilience metrics from a multivariate perspective. Finally, we support the contention that ecological stability is nested within ecological resilience: stability predicts the behavior of systems within an ecological regime, and resilience predicts the maintenance of regimes and behavior of systems collapsing into alternative regimes.