As cumulative impacts drive increased coral mortality, there is potential for higher rates of coral rubble generation and persistence. “Too much” rubble on reefs decreases resilience through impeding coral settlement and growth. Identifying which of the 3,800 reefs on the Great Barrer Reef have a high risk of rubble generation is a challenge. We developed statistical models (generalized linear model, random forest, and neural networks) to predict rubble using the GBR10 Benthic Habitat map with 10m resolution. Rubble generation and persistence factors modelled included cumulative Degree Heating Weeks (DHW), mean DHW, crown-of-thorns seastars, cyclone tracks, cyclone wave height, wave height, wave energy density, bottom velocity, surface current, ship tracks, tsunamis, bioregion, and latitude. The regression model had 70% accuracy of predicting rubble with a 28% false positive rate while the machine learning methods had higher accuracy (~90%) and lower false positives detection (~5%). However, to assess whether these tools are accurate enough, we need to consider coral reef rubble modelling from the perspective of coral management decisions. Our results suggest that probability of rubble is likely to increase in future warming scenarios. These models could be used to prioritise areas of monitoring for rubble increases and sites for intervention deployment.