We present a framework on how to estimate the costs of environmental regulations by leveraging their multiperiod design. Our application focuses on the National Ambient Air Quality Standard (NAAQS) for ozone, which was profoundly revised in 1997: A three-year running average was supposed to reduce the role of random meteorological factors that can push the ambient air quality level to cross the admissible level. Fluctuations in previous years can be counterbalanced by adjusting emissions in the current year. First, we show that daily temperatures are significant predictors of the daily maximum 8-hr ozone average upon which the standard is based. This allows to construct exogenous variations in ozone induced by temperature fluctuations. Second, we find a significant negative auto-correlation following the 1997 revision: temperature-driven ozone shocks in previous years are offset by opposite changes in ozone concentration in the current year, through reductions in NOx and VOC emissions by industrial boilers and petroleum plants. Third, we use the observed response, combined with estimates of either the non-attainment cost or the marginal abatement cost, to back-out the cost of the other type. The proposed method can be applied beyond ozone, to the broader set of multiyear pollution standards.