Estimating the value of time (VOT) in transport is primarily an empirical question, despite some strong theoretical foundations from microeconomics. The most common estimation approach has been based on discrete choice models, which produce estimates of the users' subjective VOT. These estimates are often used as inputs into estimates of user benefits for economic appraisals. The subjective and social values are thought to be roughly equal, though several authors have analyzed the possibility of divergence between these two measures. How estimates of the subjective VOT can produce biased estimates of the social VOT is analyzed. The discussion starts with a conceptual basis for possible differences between subjective and social measures. In an empirical section, stated-preference and revealed-preference data from Croatia are analyzed. The results lend support to the hypothesis that the implied subjective VOT, as revealed by mode choice data, may in some instances be a poor approximation of the social VOT, in particular when travelers are not directly bearing the opportunity cost of their travel time.