We adopt a lognormal distribution for earthquake interval times, and we use a locally determined rather than a generic coefficient of variation, to estimate the probability of occurrence of characteristic earthquakes. We extend previous methods in two ways. First, we account for the aseismic period since the last event (the “seismic drought”) in updating the parameter estimates. Second, in calculating the earthquake probability we allow for uncertainties in the mean recurrence time and its variance by averaging over their likelihood. Both extensions can strongly influence the calculated earthquake probabilities, especially for long droughts in regions with few documented earthquakes. As time passes, the recurrence time and variance estimates increase if no additional events occur, leading eventually to an affirmative answer to the question in the title. The earthquake risk estimate begins to drop when the drought exceeds the estimated recurrence time. For the Parkfield area of California, the probability of a magnitude 6 event in the next 5 years is about 34 per cent, much lower than previous estimates. Furthermore, the estimated 5-year probability will decrease with every uneventful year after 1988. For the Coachella Valley segment of the San Andreas Fault, the uncertainties are large, and we estimate the probability of a large event in the next 30 years to be 9 per cent, again much smaller than previous estimates. On the Mojave (Pallett Creek) segment the catalog includes 10 events, and the present drought is just approaching the recurrence interval, so the estimated risk is revised very little by our methods.