Most studies suggest that environmental taxes are regressive, making them less attractive policy options. We consider the distributional effects of a gasoline tax increase using four incidence measures and under three scenarios for gas tax revenue use. To incorporate behavioral responses we use Consumer Expenditure Survey data to estimate a consumer demand system that includes gasoline, other goods, and leisure. Our estimates confirm that when revenues are not recycled, a gasoline tax is regressive. Use of incidence measures that ignore demand responses, however, will substantially overstate this regressivity. In contrast, the differences between the equivalent variation and easier-to-implement consumer surplus measures are relatively small. In addition, our results suggest that using the additional gas tax revenue to fund labor tax cuts makes the policy substantially less regressive while using the revenue to fund lump-sum transfers actually makes it progressive.