The present study assessed whether factual nutritional information on portion sizes helps consumers to select healthier meals. 124 people were invited to serve themselves lunch from a 'fake food buffet' containing 55 replica food items. Participants in the control group were instructed to serve themselves a meal, as they would normally eat from the given selections (control). Participants in the second condition were asked to select a healthy, balanced meal (instruction). People in the third group were also instructed to select a healthy meal, but in addition, they received nutrition information (instruction+information). The results suggest that participants in the instruction and instruction+information condition chose fewer sweets and desserts (F((2,121))=6.91, P<.05) but more fruits (F((2,121))=5.16, P<.05). This led to overall healthier meals than in the control group. All other food categories, including vegetables, were not altered. No difference was found between the two experimental groups. The results indicate that factual nutrition information does not help consumers compose healthier meals.