Abstract The species-area relationship (SAR), i.e. the increase in species number with area, has been repeatedly used to predict species extinction, at both local and global scales, with habitat reduction. He and Hubbell (Nature, 2011; 473, 368–371), however, argued that the function that relates species loss with decreasing habitat area cannot be simply obtained reversing the species-area accumulation curve. Using a statistically more appropriate curve based on endemics (EAR), they concluded that the SAR overestimates species extinction. Although we agree that SARs and EARs have different shapes, this does not imply that SARs overestimate species extinction. Empirical evidence suggests that SARs do not overestimate, but underestimate species extinction by habitat loss and fragmentation. We discuss various examples taken from recent literature to show that SARs underestimate species extinction.