Objective: Empirical evidence of attentional bias (AB) for food in obese and restrained eaters is contradictory. It has been suggested that AB for food in obese people might reflect both food craving and food-related concerns. Thus, AB for food may be a dynamic process. Methods: A new computational methodology (Zvielli, Bernstein, & Koster, 2015) was adopted to reanalyze thereaction time (RT) and dwell time of three food-related dot-probe task with eye-tracking studies (Werthmann et al.,2015; Werthmann et al., 2011, 2013). This new computing method uses a sequence of bias scores to express the dynamic changes of AB. Moreover, the variability of RT on filler trials was also calculated. The critical groups in these studies were overweight/obese adults, obese children, and healthy-weight restrained eaters. Results: Both the variability of AB for food stimuli and the variability of RT on filler could significantly predict the variance in body mass index (BMI). When controlling for the variability of RT on filler trials and mean AB score, larger variability of AB for food stimuli still existed in obese children and aggregated dataset. The variability of AB for food stimuli demonstrated no significant correlation with restrained eating scores and dwell time variability. Conclusions: Overweight/obese individuals are characterized by more variability in attention, and this variabilityis mainly unspecific. It probably reflects less effective executive control ability.