Food handling and eating are central to the skill of primate hand movements, and their analysis can provide insights into the evolutionary origins of hand use and its generalization to other behaviors, such as tool use. Vision contributes differently to the reach, grasp, and withdraw-to-eat components of hand use when eating, suggesting that these component movements are controlled by different visuomotor networks with distinct evolutionary histories. This study examines the role of gaze in mediating the withdraw-to-eat movement in human participants eating various food items, including candy, donuts, carrots, bananas, and apples, or pantomiming the eating movements for some of these items. Eye-tracking and frame-by-frame video analyses are used to describe gaze, gaze duration, gaze disengagement, eye blinking, and hand preference in eating each food item. The results show that gaze first identifies points on a food item that the dominant hand can grasp and then identifies points on the food item that the mouth can bite. The hand and finger shaping movements of both the initial grasp and subsequent food handling aid in exposing targets on the food for grasping and biting. The comparison of real and pantomime eating suggests that only real food items possess the affordances that elicit gaze patterns associated with identifying online targets for grasps and bites. The findings are discussed in relation to idea that gaze has a feature-detector-like role linking food cues to the skilled movements of hand shaping to grasp a food item and then to orient a food item to the mouth for biting.