This paper describes a study undertaken to estimate a departure-time and mode-choice model for Stockholm. The model is segmented according to trip purpose, and a mixed - or error component - logit model is estimated. Estimation draws on stated preference data collected from drivers travelling toward the city centre during morning peak hours. The study uncovers drivers' preferences for scheduled delay, unexpected delay, travel time and cost as well the patterns of substitution between mode and time of day alternatives. The result indicates that disutility of unexpected delay depends on the scheduled deviation from preferred arrival time. The preference for scheduled delay is roughly proportional to the time shift and varies in the population, but is much more consistent within an individual. Another finding is that constraints at the destination mainly restrict late arrival, whereas constraints at the origin mainly restrict early departure.