We tested a theoretical model of personality structures underlying patterns of intra-individual variability in contextualized appraisals. The KAPA (Knowledge-and-Appraisal Personality Architecture) model was tested experimentally among smokers appraising their efficacy to resist the urge to smoke in high-risk situations. In a novel design, we assessed self-knowledge and situational beliefs idiographically and employed cognitive priming to manipulate the accessibility of self-knowledge experimentally. The results confirmed the unique KAPA-model prediction that priming would affect appraisals in a contextualized manner. Priming positively valenced self-knowledge enhanced self-efficacy appraisals specifically within that subset of situations that were relevant to the primed knowledge. The results were consistent with the hypothesis that systems of self- and situational knowledge underlie consistency and variability in appraisals.