This paper examines the effect of smoking behavior on earnings. Using data from the GSOEP, both cross-sectional and longitudinal models are estimated separately for males and females. Results for the cross-sectional models confirm prior analyses inasmuch as smoking has a negative effect on earnings for males. However, applying fixed-effects estimation, this effect is found to be inverted for men aged 25 to 35 years compared to their non-smoking counterparts. That is, controlling for unobservable individual heterogeneity, the result implies that male smokers are individuals with higher time preference rates. At the early stage of the age-earnings course higher earnings are therefore found for smokers because young male non-smokers only are about to start off their occupational career. Women?s earnings, however, are not affected by smoking behavior.