Several studies have documented the effectiveness of bupropion for smoking cessation, yet little is known about the mechanisms by which it facilitates abstinence. In this placebo-controlled randomized trial. We examined whether bupropion's effects on cessation were mediated by changes in withdrawal and/or negative or positive affect (PA). Two hundred and fifty-one smokers received 10-week treatment with bupropion or placebo, plus behavioral counseling. Changes in affect and withdrawal symptoms from pre-quit to 1 week post quit were examined as mediating variables in structural equation models. Cotinine-verified 7-day point prevalence cessation rates at the end of treatment (8-weeks post quit date) were 48% for bupropion and 29% for placebo (P=0.001). There were significant treatment effects on withdrawal and negative affect (NA); however, only change in NA predicted cessation. In a path model, change in NA was a significant mediator of bupropion's effects on cessation. However, the proportion of variance accounted for by this mediator was small, suggesting that other unmeasured factors play an important role. Laboratory-based paradigms may be useful to identify other mediators of bupropion's effects, thereby pointing to mechanisms of effect that can be bolstered in future treatment studies.