Prospective data from the California Tobacco Surveys (n = 2066) were used to perform a critical test of the Prochaska et al. (1991) stages of change model. When the stages of change model was used as a stand alone predictor, smokers in preparation at baseline were more likely to be in cessation at follow-up than smokers in pre-contemplation at baseline (ORadj = 1.9). When stage membership was combined with baseline measures of addiction including smoking behaviors and quitting history, it was not a significant predictor of future cessation. A prediction equation that combined daily vs. occasional smoking, cigarettes per day smoked, life-time quits of at least a year, and quits of more than 5 days in the previous year discriminated smokers in cessation at follow-up of 1 to 2 years better than did the stages of change model. The area under the ROC curve for the equation based on addiction measures was 69.3% vs. 55.1% for the stages of change. Cessation rates ranged from 7.7% to 35.7% for the four-category addiction equation compared with 15.1% to 24.9% for stages of change model.