In order to test the effect of calorie information on fast food choices, we conducted a questionnaire employing two types of stated preferences methods (the best-worst-scaling and intentional questions) and a follow-up randomized field experiment in a sample of 119 participants. This combined approach allowed us to test the internal validity of preferences for fast food meals across elicitation scenarios. The results showed that calorie information reduces the probability of selecting high calorie meals only in the questionnaire, while it did not have any significant impact on actual purchasing behavior in the field experiment. Thus, the findings show that there is a clear difference between the role of calorie information on immediate stated preference choices, and the relatively low level of responsiveness in real choices in a restaurant. We believe that the current results are quite suggestive, indicating the limits of predicting actual fast food behavior, and may open the way to using data sources that combine stated methods with field experiments.