Previous studies of transport tax reform have typically assumed that the reform itself does not affect the marginal value of time. In this paper we consider a model of urban transport with two trip purposes, commuting and non-commuting, to analyse the effects of transport tax reform on the value of time and marginal external congestion costs. The theoretical results suggest that the assumption of multiple trip purposes implies that these effects are non-trivial. Consequently, assuming exogenous time values may lead to inaccurate estimates of optimal congestion taxes and of the welfare effects of transport tax reform. Empirical work using Belgian data illustrates the potentially large effect of transport tax reform on time values. In fact, the majority of the tax reform exercises studied reduce traffic levels but raise time values and marginal external congestion costs.