Abstract. Generalized anxiety disorder (GAD) is associated with worry and emotion regulation difficulties. The contrast-avoidance model suggests that individuals with GAD use worry to regulate emotion: by worrying, they maintain a constant state of negative affect (NA), avoiding a feared sudden shift into NA. We tested an extension of this model to positive affect (PA). During a week-long ecological momentary assessment (EMA) period, 96 undergraduates with a GAD analog provided four daily measurements of worry, dampening (i.e., PA suppression), and PA. We hypothesized a time-lagged mediation relationship in which higher worry predicts later dampening, and dampening predicts subsequently lower PA. A lag-2 structural equation model was fit to the group-aggregated data and to each individual time-series to test this hypothesis. Although worry and PA were negatively correlated in 87 participants, our model was not supported at the nomothetic level. However, idiographically, our model was well-fit for about a third (38.5%) of participants. We then used automatic search as an idiographic exploratory procedure to detect other time-lagged relationships between these constructs. While 46 individuals exhibited some cross-lagged relationships, no clear pattern emerged across participants. An alternative hypothesis about the speed of the relationship between variables is discussed using contemporaneous correlations of worry, dampening, and PA. Findings suggest heterogeneity in the function of worry as a regulatory strategy, and the importance of temporal scale for detection of time-lagged effects.