It is now clear that positive emotion leads to enhanced psychological functioning. What is less clear, however, is just why this is so. Drawing on a social-functional perspective, we argue that positive emotional behavior that accurately signals to others the individual's internal state will enhance social connectedness. Positive emotional behavior that does not accurately signal a person's experience--such as a smile that is not felt--may impede social connectedness and, in turn, psychological functioning. This perspective suggests that (a) the degree to which experience and behavior are dissociated during positive emotional episodes, over and above level of positive behavior, should predict worse psychological functioning and (b) the effect of dissociation should be mediated by social connectedness. To test these hypotheses, we conducted a short-term prospective longitudinal study, with a baseline assessment of depressive symptoms and well-being at Time 1. Six months later, at Time 2, we used a novel within-individual laboratory paradigm to measure the degree to which positive emotional behavior was dissociated from (vs. coherent with) a participant's positive emotional experience. We also assessed level of positive behavior and experience. Then, another 6 months later, we assessed social connectedness as a mediator and depressive symptoms and well-being as outcomes at Time 3. Even when controlling for baseline functioning and for level of positive emotion behavior and experience, we found that greater positive experience-behavior dissociation at Time 2 predicted higher levels of depressive symptoms and lower levels of well-being at Time 3. As predicted, these associations were mediated by social connectedness.