The current doctoral thesis sought to develop an IRAP that could assess obese and normal-weight individuals’ attitudes to healthy and unhealthy foods. Three empirical studies directly compared the ability of IRAP and explicit measures to assess obese and normal-weight individuals’ food biases in a two-hour and unrestricted food deprivation state. An additional objective of the research programme was to determine if it was possible to detect reliable differences in neurophysiological activity while participants completed a food-attitude IRAP. Finally, the research aimed to examine the malleability of implicit attitudes to healthy and unhealthy foods. All studies presented participants with an IRAP and explicit measures. The IRAP presented “pro-unhealthy” and “pro-healthy” trials. The difference in mean-responselatency between “pro-healthy” and “pro-unhealthy” trials indicated participants’ bias towards healthy or unhealthy foods. The advantages of the IRAP were highlighted across the empirical investigations: (a) unlike any other implicit measure, it differentiated between the implicit responses of obese and normal-weight individuals to healthy and unhealthy foods, accounting for variance beyond that provided by a range of explicit measures; (b) the IRAP effects were relatively robust across studies; (c) a measure of neurological processing (EEGs) was successfully obtained while participants completed the IRAP, and the findings yielded some patterns that appear consistent with previous research; and (d) it revealed the malleability of implicit responses using an acceptance-based intervention, an effect that has not yet been reported in the literature on psychological acceptance or implicit attitudes. Overall, therefore the pattern of results in these studies highlighted the utility of the IRAP for future investigations of implicit food attitudes among obese and normal-weight individuals. Finally, the current research programme adds to previously published IRAP studies showing the efficacy of the IRAP as a measure of implicit bias across a range of domains.