Highlights • Uncertainty is ignored in virtually all carnivore studies of compensatory mortality. • We tested for compensation and additivity in 2 cougar populations, >16 years of data. • We accounted for uncertainty, density dependence and heterogeneity in mortality risks. • We could not reject the additivity hypothesis in the heavily harvested population. • We detected partial compensation in the semi-protected population ignoring uncertainty systematically led to biased conclusions.