Although reliability of travel modes has generally been found to be one of the most important attributes of transportation systems, few attempts have been made to make these findings operational. The impact of a specific measure of reliability, the variability of travel time, on the route-to-work choice is investigated in this research through the use of a survey instrument posing hypothetical commute alternatives. A mean-variance model of travel choice is used to estimate an individual's tradeoff between time and variability (i.e., risk aversion). The principal tool of analysis was LINMAP, a linear programming technique for multidimensional analysis of preferences, which yielded estimates of each respondent's risk aversion coefficient (λ). A pretest was conducted to determine an appropriate format for the commute choices and to investigate external validity concerns.