Modeling of departure time choice has recently received renewed attention because of the increasing levels of congestion in many cities and the growing popularity of travel demand management strategies such as road pricing. Current practice for evaluation of the effectiveness of travel demand management policies usually involves incorporation of the temporal dimension into transport planning models only through fixed factors derived from origin-destination data, but this practice makes them unsuitable for use in proper prediction of demand at different times of the day. To mitigate these deficiencies, the authors argue in favor of estimation and application of specially formulated time-of-day choice models. Here the concentration is on the process used to generate a survey design that obtains data suitable for estimation of such models that will ensure both realism and simplicity in the presentation; in particular, the stated preference exercise includes dependence between attribute levels. The proposed procedure should be widely applicable and offers many improvements over current practice in the field.