Relapse is a central problem in smoking treatment. Data collected at the time of relapse episodes indicate that stress and negative affect (NA) promote relapse, but retrospective data are potentially biased. The authors performed a prospective analysis of stress and NA prior to initial lapses in smokers (N = 215). Day-to-day changes in stress (daily negative and positive events and Perceived Stress Scale scores) and NA (multiple momentary affect ratings) did not predict lapse risk on the following day. However, within the lapse day itself, NA was already significantly increasing hours before lapses, but only for episodes attributed to stress or bad mood. Thus, rapid increases in NA, but not slow-changing shifts in stress and NA, were associated with relapse.