The travel cost method is widely used to estimate the value of recreational sites and to simulate the impact of public policies on welfare. Congestion is likely to be a determinant of recreational demand (Jakus and Shaw, 1997) and has to be given special attention. Ignoring this variable leads to the omission bias. But introducing congestion is also likely to involve endogeneity problems because of a potential correlation with the error term. In addition, congestion is determined by a sorting equilibrium, i.e., it is both the result of the choice process and an explanatory variable of this choice. Some methods have been recently propose to address these difficulties. These methods are presented and illustrated in a nested model of recreational choice.