Research has suggested that the time course of the smoking withdrawal syndrome is fairly invariant across smokers and that smoking withdrawal symptoms are weakly related to relapse. Withdrawal data from 2 clinical trials of the nicotine patch were analyzed to evaluate these characterizations. In both studies, patients were clustered according to the shapes of their withdrawal profiles across 8 weeks of treatment. In each study, 3 clusters with distinct temporal patterns of withdrawal symptomatology emerged. Clusters included both abstinent and lapsing patients, and patch dose was unrelated to cluster membership. Patients with "atypical" patterns of smoking withdrawal (e.g., late symptomatic elevations) were more likely to relapse than patients who showed a gradual elimination of withdrawal. Withdrawal shape, duration, and severity all contributed significantly to the prediction of relapse. Measures of negative affect closely tracked withdrawal symptoms over time within clusters. Topics for future smoking withdrawal research are discussed.