This study compares ex ante estimates of the direct costs of individual regulations to ex post assessments of the same regulations. Our review of more than two dozen environmental and occupational safety regulations indicates that ex ante estimates of total (direct) costs have tended to exceed actuals. We find this to be true of 12 of the 25 rules in our data set, while for only 6 were the ex ante estimates too low. The overestimation of total costs is often due to errors in the quantity of emission reductions achieved by the rule which, in turn, suggest that the rule's benefits may also be overestimated. The quantity errors are driven by both baseline and compliance issues. At least for EPA and OSHA rules, overestimation of per-unit abatement costs occurs about as often as underestimation. In contrast, for those rules that use economic incentives, per-unit costs are consistently overestimated. Much of the overestimation can be attributed to technical innovations unanticipated at the time the rule is issued, and to quantity errors. In addition, several methodological and procedural explanations also apply: changes in the regulation after the cost estimate is prepared, use of maximum cost estimates, and asymmetric error correction. Since a number of environmental laws encourage the development of cost estimates that reflect a maximum rather than a mean, regulatory agencies could issue a "best estimate" along with the statutorily preferred cost estimate. Likewise, they could ensure that rule changes made in the course of the regulatory development process are manifest in revised cost estimates. Indeed, discovering how and when to adjust ex ante estimates provides the strongest possible justification for more credible ex post studies--a research activity that merits greater emphasis.