A common finding in stated preference studies that measure the value of travel time (VTT), is that the measured per-minute VTT increases with the size of the time change considered, in conflict with standard neoclassical theory. The current paper tests prospect theory as a possible explanation: More specifically, whether the phenomenon is generated by preferences being reference-dependent and exhibiting diminishing sensitivity for gains and losses, with a stronger degree of diminishing sensitivity for money than for travel time. We use stated preference data with trade-offs between travel time and money that pro- vide identification of the degrees of diminishing sensitivity for time and money gains and losses, thus enabling us to test and potentially falsify the prospect theory explanation. We apply a discrete choice model, in which choice depends on a reference-free value of travel time and reference-dependent value functions for time and money, allowing for loss aver- sion and different degrees of diminishing sensitivity for gains and losses. We use semi- parametric local logit estimates of the equi-probability curves in the data to test the model’s appropriateness, and estimate its parameters using a mixed logit approach. Our results sup- port the prospect theory explanation.