The recent elevated rate of large earthquakes has fueled concern that the underlying global rate of earthquake activity has
increased, which would have important implications for assessments of seismic hazard and our understanding of how faults interact. We examine the timing of large (magnitude M ≥ 7) earthquakes from 1900 to the present, after removing local clustering related to aftershocks. The global rate of M ≥ 8 earthquakes has been at a record high roughly since 2004, but rates have been almost as high before, and the rate of smaller earthquakes is close to its historical average. Some features of the global catalog are improbable in retrospect, but so are some features of most random sequences—if the features are selected after looking at the data. For a variety of magnitude cutoffs and three statistical tests, the global catalog, with local clusters removed, is not distinguishable from a homogeneous Poisson process. Moreover, no plausible physical mechanism predicts real changes in the underlying global rate of large events. Together these facts suggest that the global risk of large earthquakes is no higher today than it has been in the past