Transportation authorities have consistently failed to employ economic incentives on major roadways--i.e. time-of-day pricing or "congestion fees"--to internalize the costs of congestion. In principle at least, such tolls can easily be shown to increase social welfare by making motorists pay something closer to the full social costs of their driving decisions. In addition, recent advances in electronics make it possible to implement such fees fairly cheaply and non-intrusively. While these same authorities generally understand and acknowledge the case for using congestion fees, they also claim that their use is politically infeasible because too many motorists would suffer large increases in commuting costs. This is the puzzle: If congestion tolls truly do advance social welfare, why is it so difficult to find a way to implement them? Two common explanations for this difficulty are the following: (i) The public perceives, or politicians fear that they would perceive, such fees simply as tax increases. If so, they might be responsive to an explicit promise to return the money in some way. (ii) Motorists dislike congestion fees because they find them coercive, in that they often have few if any practical alternatives to paying the fee. If so, then a policy option that offers motorists a choice of toll lanes and the more customary free lanes may be more attractive than a policy that policy of tolls on all lanes.</p><p>We have completed a survey of Southern California residents designed to test these hypotheses. Unlike most opinion surveys on congestion pricing, our survey was quite explicit about the fate of the collected revenues. For example, we presented respondents with policies that returned a substantial portion of the revenues to the public, either in the form of cash (through reductions in sales taxes and vehicle registration fees or through income tax credits) or in the form of coupons to be used for vehicle emissions equipment repair, transit, and the like. In the past, most surveys have not been explicit about the revenues, or they have stated that revenue use was to be for improved highways, transit, or other public purposes. We find that a promise to offset the imposition of congestion fees by other taxes can result in a 7 percentage point increase in support for congestion pricing policies, and the restriction of congestion pricing to a single lane on a freeway attracts from 9 to 17 percentage points of additional support.