This paper examines the proposition that stress shrinks affective space, increasing the inverse correlation between positive affect (PA) and negative affect (NA). The experience sampling method was used to record the levels of PA and NA and stress reported by white-collar employees 10 times a day for 5 days. These data were subjected to hierarchical linear modelling to determine whether the relationship between affective states becomes increasingly inverse as a function of stress, as predicted by Zautra, Potter, and Reich (1997). Caution was taken to address measurement issues that have been raised in recent debates over the independence of PA and NA, and a contingency analysis was also used to supplement the linear modelling approach. Both types of analyses revealed evidence consistent with the hypothesis that the degree to which PA and NA are inversely related varies with the level of stress.