This paper addresses the problem of measuring the welfare benefits of a transport improvement. We formulate and analyze a rich spatial model that allows for spillovers, matching and income tax, in a setting with multiple work and residential locations and very general worker heterogeneity. The conventional consumer surplus captures part of the benefits and is calculated based on predictions of changes in travel demand and transport costs. The issue is to determine which so-called wider impacts to add to this. We find that adding the change in total output as a wider impact leads to double-counting of benefits. The output change due to spillovers should be added, while the output change due to matching is already partly included in the consumer surplus. These results are useful for applied cost-benefit analysis of transport policies