There is a broad consensus that the rise in the obesity epidemic in the past several decades has been driven more by overeating (increased calorie intake) than by expending fewer calories (Ludwig & Nestle, 2008; Swinburn, et al., 2011). These effects are compounded by today’s oversized food portions (Chandon & Wansink, 2007; Rolls, Roe, & Meengs, 2007). Supersized portions increase food intake partly because people rarely read the size information on the labels and instead rely on their visual impressions to estimate portion size. However, visual perception is biased, and consumers tend to underestimate the changes in food portions, more so for large portions than for small portions. As a result, consumers overeat because they fail to realize just how large supersized portions really are (Chandon & Wansink, 2007). These findings have brought the biases in portion size perception and their remedies to the forefront of the research and the policy agenda. Previous research on this issue focused on the effects of physical properties of packages such as size and shape, and on solutions to design packages that do not bias visual perception (Chandon & Ordabayeva, 2009; Folkes & Matta, 2004; Raghubir & Krishna, 1999; Wansink & Van Ittersum, 2003). In this research, we examine how higher-order factors, namely, ambivalent attitudes towards food (perceiving the food as desirable but harmful) influence size perceptions. Our results propose strategies that can effectively reduce biases in portion size perception without requiring any changes in physical package attributes such as size or shape (which may be costly, time consuming, and difficult to enforce through policy).