Abstract Historical biogeography seeks to understand the distribution of biodiversity in space and time. The dispersal-extinction-cladogenesis (DEC) model, a likelihood-based model of geographic range evolution, is widely used in assessing the biogeography of clades. Robust inference of dispersal and local extinction parameters is crucial for biogeographic inference, and yet a major caveat to its use is that the DEC model severely underestimates local extinction. We suggest that this is mainly due to the way in which the model is constructed to allow observed species to transition into being present in no areas (i.e., null range). By prohibiting transitions into the null range in the transition rate matrix, we were able to better infer local extinction and support this with simulations. This modified model, DEC*, has higher model fit and model adequacy than DEC, suggesting this modification should be considered for DEC and other models of geographic range evolution.