The unidimensionality of nicotine dependence has not been established firmly yet. The aim of this study was to assess the dimensionality of nicotine dependence, preferably meeting the strict assumptions of the Rasch model. First, we examined the validity of the Fagerström Test for Nicotine Dependence (FTND) [Br. J. Addict. 86 (1991) 1119.] in 1525 smokers who participated in a national survey considering smoking behavior. Two factors were found, suggesting that the FTND does not measure a unidimensional construct. Factor analysis of 19 other dependence items in 512 smokers resulted in four factors of which three were interpretable: compulsive smoking, social problems due to smoking, and physical dependence. We focused on smoking compulsivity. This factor turned out to consist of a four-item Rasch homogeneous scale. Two items of the FTND with face validity of smoking compulsivity were found to fit into the scale. The results of Rasch analysis were in support of a continuum of compulsivity. Difficulty refraining from smoking in places where it is forbidden was found to indicate highest compulsivity. Several correlates with smoking compulsivity were found. We conclude that compulsive smoking is one important dimension of nicotine dependence, which may account for the considerable relapse of this disorder.