Research into rape-supportive cognition has been hampered by methodological problems such as social desirability. The study of so-called implicit cognition has been dominated by explanations that the effects produced on implicit/indirect measures are mediated by associations held in memory. The Implicit Relational Assessment Procedure (IRAP; Barnes-Holmes et al., 2006) offers an alternative perspective. Derived from Relational Frame Theory (RFT, Hayes, Barnes-Holmes & Roche, 2001), here it is applied for the first time to study the malleability of implicit rape-supportive cognition, following a cognitive-restructuring task in university males. The relationship between implicit (IRAP) and explicit measures (the Acceptance of Modern Myths about Sexual Aggression scale (AMMSA) Gerger, Kley, Bohner & Siebler, 2007), and behaviour was explored (measured by a forced-choice task and ratings of the researcher). IRAP scores did not change following the intervention. Behaviour on the forced-choice measure was predicted by IRAP scores but not by scores on the AMMSA. Additional analyses into the predictive ability of the AMMSA and IRAP measure on behavioural measures (charity-box overall giving score and Researcher Rating Scale) were conducted. Findings from hierarchical multiple regression analyses indicated that the AMMSA predicted none of the variance whereas the IRAP predicted 12.4% and 11.5% of the variance respectively.