Although bupropion is known to be an effective aid to smoking cessation, little is known about its mode of action. In the present study we tested the hypothesis that bupropion reduces the likelihood that a smoking lapse, or slip, leads to a subsequent relapse. This hypothesis was tested in the context of a clinical trial of bupropion as a smoking cessation aid, using Cox regression and representing lapse history as a discrete time-varying covariate. Bupropion treatment reduced the probability of relapse during the treatment phase (hazard ratio, or HR=.421, p< or =.000) but not during the follow-up phase (end of treatment to 6 months, HR=.896, p< or=.67). As anticipated, having small lapses during treatment contributed to or predicted subsequent relapse, both during treatment (HR=2.897, p< or =.000) and during follow-up (HR=2.320, p< or=.008). Although an interaction was found between drug treatment and lapse history in predicting subsequent failure during the treatment phase, the finding suggested that drug slightly increased the effect of lapse on eventual failure during treatment (HR=1.706, p<or=.012). No evidence was found that the effect of this interaction carried forward into follow-up (HR=1.500, p<or=.30). Although the mechanisms of action of bupropion may best be addressed in a human laboratory study, results from this clinical trial suggest that the effectiveness of bupropion treatment is not attributable to reducing the effect of a lapse in predicting relapse.