Although smuggled cigarettes have been a prevalent problem and a severe challenge to public health and welfare around the world, little is known about the behavior associated with smoking smuggled cigarettes and the issue is difficult to study due to data limitations. By means of a population‐based tobacco survey conducted in Taiwan, the present paper applies a latent class model to identify potential smokers who are either currently or will at some point in the future be consuming smuggled cigarettes. This methodology, in contrast to the traditional discrete models, allows potential smokers who are more inclined to smoke smuggled cigarettes to be endogenously classified. The empirical results indicate that socio‐demographic factors do increase the inclination to smoke smuggled cigarettes after unobserved heterogeneity has been accounted for.