This study contributes to the understanding of the decision to quit sm
oking by taking into account the learning of new risk information. The
specific hypothesis tested is that smokers learn new risk information
and hence create an incentive to quit from their own experience. Prob
it models are estimated for the decision to quit smoking based on long
itudinal data obtained from Taiwan. It is shown that health risk, meas
ured by the observed change in health status over the period between t
wo surveys, has a relatively substantial positive effect on the probab
ility of quitting smoking. In addition, the results indicate that scho
oling has a significantly positive effect on the probability of quitti
ng. These findings are consistent with the predictions of a Bayesian l
earning framework and suggest that the risk information obtained from
individual experience, which is the sole source of information availab
le to smokers in most developing countries, plays the same role that p
ublic information does.