The minimum-cost travel time data-collection problem is analogous to a vehicle routing problem with pickup and delivery, subject to the constraints of time windows, because the problem involves making decisions regarding two aspects: vehicle routing and vehicle scheduling. The proposed bilevel programming model minimizes the total travel cost of collecting travel time data with probe vehicles. The upper-level model is designed to determine fleet size and feasible schedules, whereas the lower-level model attempts to route vehicles at the least cost. The model in the lower level consists of a total cost objective function and a set of practical constraints, such as number of probe vehicles, number of study roadway segments, and project duration. Travel time data collection on a highway network containing several corridors in New Jersey was applied to demonstrate the applicability and performance of the developed model.