The goal of this study is to examine individual differences in the use of affective and cognitive information when forming attitudes, or affect-cognition reliance (ACR). An empirical measure of this construct is developed through the use of latent class modeling and multilevel modeling. The proposed measure of ACR is compared with existing personality measures and validated in an applied context by focusing on judgments of job applicants through resume ratings. Results indicate that the multilevel measure of ACR has a coherent nomological network in terms of convergent validity with personality measures and also interacts with resume quality to predict resume ratings, which bears implications for real world hiring situations.