The paper uses an applied general equilibrium model, calibrated to the situation in Belgium in 1990, to evaluate the welfare effects of small policy changes in the presence of transport externalities. The model incorporates three types of externalities: congestion, which has a feedback effect on the behaviour of the economic agents, air pollution and accidents. The model is used to perform balanced budget incidence simulations in which the marginal cost of public funds is calculated for four alternative policy instruments: a lump sum tax, the labour income tax, the fuel taxes and peak road pricing. For each of these instruments the marginal cost of public funds (MCF) is calculated. The results of the model are compared with those of a model in which congestion, air pollution and accidents are assumed to remain constant at their initial level. The information provided by this type of exercises is useful for two purposes. First of all, it serves as an input in cost-benefit analyses. The non-tax costs of a public project should be multiplied by the MCF of the instrument used to finance it, in order to make a correct comparison with its benefits. Secondly, the MCF is a useful tool in deciding whether a revenue neutral policy reform is welfare improving or not. From the theory of optimal taxation in the presence of externalities, we know that when taxes are set optimally, the MCF is equalized across all tax instruments. Starting from a non-optimal tax system, social welfare is increased (decreased) when the tax with the highest welfare cost per additional unit of government revenue is reduced (increased) and when simultaneously the tax with the lowest welfare cost per additional unit of government revenue is raised (reduced). The paper contributes to the literature in two ways. First of all, it includes non-identical individuals. Therefore, the evaluation of the policy instruments takes into account the equity effects of the policy reforms. The second contribution is related to the way in which the externalities are modelled: the feedback effect of congestion is explicitly taken into account and the value of a marginal time saving is determined endogenously in the model. The structure of the paper is as follows. After the introduction, we first describe the general characteristics and the specification of the applied general equilibrium model, with a particular attention to the modelling of the externalities. It is followed by a brief description of the initial equilibrium, which corresponds with the situation in Belgium in 1990. Next, we discuss the results of the balanced budget incidence simulations. The final section concludes and describes some extensions to the model. The simulations show that the ranking of the instruments in terms of their marginal cost of public funds changes significantly when the effect of the reform on the externalities is taken into account. Secondly, regardless of the way in which the tax revenue is recycled, the welfare gain of peak road pricing is higher than that of the fuel tax. When the externality tax revenue is recycled through the lump sum tax the welfare gains are higher for the poorer than for the richer quintiles. On the other hand, the main beneficiary of revenue recycling through the labour income tax is the richest quintile. Consequently, when the social welfare function gives a higher weight to the welfare of individuals belonging to the poorer quintiles, the distributional impacts of the policy reforms cause the welfare gain to be higher when the revenue is recycled through an increase in the lump sum transfer rather than through a lower labour income tax rate.