Decisions on the merit of strengthening buildings to resist earthquakes should not be made on a cost benefit basis, and in particular should not use estimates of the benefit that are based on the average rate of occurrence of strong ground motion. The distribution of expected ground motion is so skewed that no central measure provides a good representation of the risk. A better procedure is to determine what is the unacceptable level of loss, and then to engineer to the strength that will prevent that loss.