It has been argued that obese individuals evaluate high caloric, palatable foods more positively than their normal weight peers, and that this positivity bias causes them to consume such foods, even when healthy alternatives are available. Yet when self-reported and automatic food preferences are assessed no such evaluative biases tend to emerge. We argue that situational (food deprivation) and methodological factors may explain why implicit measures often fail to discriminate between the food-evaluations of these two groups. Across three studies we manipulated the food deprivation state of clinically obese and normal-weight participants and then exposed them to an indirect procedure (IRAP) and self-report questionnaires. We found that automatic food-related cognition was moderated by a person's weight status and food deprivation state. Our findings suggest that the diagnostic and predictive value of implicit measures may be increased when (a) situational moderators are taken into consideration and (b) we pay greater attention to the different ways in which people automatically relate rather than simply categorize food stimuli.