We estimate a residential sorting model using microdata to study an expansion of the Vancouver rapid transit network. Our results indicate that many aspects of household preferences were stable over time; however, preferences for transit access became more homogeneous with respect to income. Simulations show that these preference changes, along with endogenous housing appreciation and rising income levels in connected neighborhoods, result in larger benefits for higher-income households. This distributional pattern is driven by the gains of households working in pre-connected neighborhoods. In contrast, the benefits to those working in newly connected neighborhoods disproportionately favor lower-income households.