Based partly on Weiss and Cropanzano's (H. M. Weiss & R. Cropanzano, 1996) Affective Events Theory, hypotheses were stated about likely antecedents and consequences of positive and negative real-time affective reactions at work. Somewhat different antecedents and consequences were predicted for positive as opposed to negative affective reactions. Affective reactions were operationalized as the average of up to 50 reports of momentary positive and negative feelings collected at work over a 2-week period. Structural equation analyses suggest that the data are consistent with the theoretical model proposed. As expected, job characteristics and positive dispositional affectivity predict positive affective reactions, role conflict and negative affectivity predict negative affective reactions, positive affective reactions predict affective commitment and helping behavior, and intention to leave is predicted by attitudes rather than by affective reactions. The results are consistent with Affective Events Theory as well as with 2-domain models that propose different causes and consequences of positive versus negative affect.