We compare the historical record of earthquake hazard experienced at 78 towns and cities (sites) distributed across New Zealand and the continental United States with the hazard estimated from the national probabilistic seismic-hazard (PSH) models for the two countries. The two PSH models are constructed with similar methodologies and data. Our comparisons show a tendency for the PSH models to slightly exceed the historical hazard in New Zealand and westernmost continental United States interplate regions, but show lower hazard than that of the historical record in the continental United States intraplate region. Factors such as non-Poissonian behavior, parameterization of active fault data in the PSH calculations, and uncertainties in estimation of ground-motion levels from historical felt intensity data for the interplate regions may have led to the higher-than-historical levels of hazard at the interplate sites. In contrast, the less-than-historical hazard for the remaining continental United States (intraplate) sites may be largely due to site conditions not having been considered at the intraplate sites, and uncertainties in correlating ground-motion levels to historical felt intensities. The study also highlights the importance of evaluating PSH models at more than one region, because the conclusions reached on the basis of a solely interplate or intraplate study would be very different.