Patient characteristics may partly explain individual differences in long-term maintenance of treatment gains in response to cognitive and behavioral therapies (CBT). This study evaluated the relative contribution of negative cognitive style, behavioral activation, and personality patterns, measured at the end of treatment among treatment responders, towards the risk of relapse of major depressive disorder over a 2-year prospective follow-up. The sample included depressed adults who were randomized and received an adequate dose of one of three forms of CBT, who then fully responded and completed a 2-year prospective follow-up (N = 93) (Jacobson et al., 1996; Gortner, Gollan, Dobson and Jacobson, 1998). Cox regression analyses revealed that for treatment responders, self-devaluative depressogenic cognitive style did not predict relapse during the 2-year follow-up. Separate cox regression analyses indicated that without considering the contribution of personality patterns, it was important to distinguish the level of behavioral activation (e.g. frequency of pleasant events) from its associated pleasure in predicting relapse. With the introduction of personality patterns; however, an integrated model revealed that only dependent personality pattern was associated with hazard of relapse during the 2-year prospective follow-up. Dependent personality pattern may increase relapse through behavioral passivity.