Pressure on natural communities from human activities continues to increase. Even unique ecosystems like the Great Barrier Reef (GBR), that until recently were considered near-pristine and well-protected, are showing signs of rapid degradation. We collated recent (1996–2006) spatiotemporal relationships between benthic community composition on the GBR and environmental variables (ocean temperature and local threats resulting from human activity). We built multivariate models of the effects of these variables on short-term dynamics, and developed an analytical approach to study their long-term consequences. We used this approach to study the effects of ocean warming under different levels of local threat. Observed short-term changes in benthic community structure (e.g., declining coral cover) were associated with ocean temperature (warming) and local threats. Our model projected that, in the long term, coral cover of less than 10% was not implausible. With increasing temperature and/or local threats, corals were initially replaced by sponges, gorgonians, and other taxa, with an eventual moderately high probability of domination (>50%) by macroalgae when temperature increase was greatest (e.g., 3.5°C of warming). Our approach to modeling community dynamics, based on multivariate statistical models, enabled us to project how environmental change (and thus local and international policy decisions) will influence the future state of coral reefs. The same approach could be applied to other systems for which time series of ecological and environmental variables are available.