The purpose of this study was to test a 2-factor model of affective health in women with rheumatoid arthritis (RA; n = 82) or osteoarthritis (OA; n = 88). Positive and negative social interactions and affect were assessed for 11 consecutive weeks. For each participant, Vulnerability and Resilience factors were created from factor analyses of positive and negative personal characteristics, respectively. Multilevel analyses tested the hypothesis that weekly changes in social interactions or affect would only be predicted by the factor of the same valence. The Vulnerability (and not the Resilience) factor predicted changes in negative interactions. The Resilience (and not the Vulnerability) factor predicted changes in positive interactions. The Vulnerability (and not the Resilience) factor predicted changes in current and next-week negative affect. The Resilience and Vulnerability factors each predicted changes in current and next-week positive affect, although the effects for Vulnerability were smaller than for Resilience. Finally, the Vulnerability factor interacted with pain to predict more future negative affect. The main implication is that both Vulnerability and Resilience should be considered in theory, research, and interventions.