This paper uses non-parametric techniques to estimate the variation in the value of travel time. Stated choice data are used for an empirical analysis in the context of a binary route choice between an expressway route and a non-expressway route. The data were collected in a pivot-style survey in Japan. Local constant regression and local linear regression are used for the model estimation. The empirical analysis shows that the estimates based on the local linear regression behave well as a cumulative function, while those using the local constant regression do not. Then the variations in the value of travel time are compared among different subgroups in the dataset. The results showed that individuals in their forties and fifties place a higher value on travel time, as do those whose household earns a higher annual income and those who spend more time traveling.