This paper investigates gender differences in smoking behavior using data from the German Socio-economic Panel (SOEP). We develop a Blinder-Oaxaca decomposition method for count data models which allows to isolate the part of the gender differential in the number of cigarettes daily smoked that can be explained by differences in observable characteristics from the part attributable to differences in coefficients. Our results reveal that the major part of the gender smoking differential is attributable to differences in coefficients indicating substantial differences in the smoking behavior between men and women rather than differences in characteristics.