Using field experiment data, we compare the effectiveness of calorie labels to a “fat tax” at reducing calories ordered. Results from a structural model of consumer demand show that numeric labels did not influence food choice, but symbolic traffic light labels caused restaurant patrons to select lower‐calorie menu items; thus, adding a traffic light symbol could enhance the effectiveness of the numeric calorie label (as currently proposed by the Food and Drug Administration). Additionally, our model projects that labels can both reduce intake more than a 10% tax on high‐calorie items and a 10% subsidy on low‐calorie items. (JEL Q18, I18)