The Chinese earthquake catalog is used to evaluate the adequacy of simple methods for calculating seismic hazards. These simple methods use a time-stationary model for seismicity and an exponential distribution for earthquake size. Earthquake occurrences in five provinces of North China (the area with the most complete history) during time segments of three lengths (50, 100, and 200 years) are used as input to the hazard analysis. The probabilities of felt shaking in 62 cities in North China are calculated for the 50-year period following each time segment, and are compared to the observed occurrences of felt shaking during the 50-year period. Data intervals of 50 and 100 years suffice for accurate estimation of probabilities of felt shaking; 200 years of earthquake history lead to poorer estimates of these probabilities if the rate of earthquake occurrence is averaged over the entire 200 years. This inaccuracy results from the apparent periodicity of activity in North China, which has a cycle of about 300 years. The implication is that, at a specific time, the most recent seismic activity is the best data base to use for calculation of probabilities of felt shaking in the near future.