This paper examined whether international variations in absolute and relative gender differences in mortality are related to the overall mortality rates, and whether the international variation in gender gap in mortality can in part be explained by smoking. I used data on mortality from all causes in 32 European countries published by the World Health Organisation, and indirect estimates of mortality attributable to smoking in the age band 35-69 years by Peto et al. The main analyses were restricted to the age band 35-69 years but results for mortality at all ages were virtually identical. The overall mortality rates (both sexes combined) were strongly related to absolute gender differences (r = 0.91) but only weakly to relative differences (r = 0.35). The gender gap was larger in eastern than in western Europe for rate differences (1005 vs. 530 per 100,000, respectively), but it was similar for rate ratios (2.3 vs. 2.1, respectively). Both absolute and relative gender differences in mortality were strongly related to the difference between men and women in the proportion of all deaths attributed to tobacco (partial correlations, after controlling for the overall death rates, were 0.59 and 0.66, respectively). Excluding tobacco-related deaths attenuated the associations between the overall mortality rates with absolute differences (r = 0.70) and reduced the difference in the absolute gender gap between eastern and western Europe. More importantly, excluding tobacco-related deaths eliminated entirely the association with relative differences (r = -0.15) as well as any suggestion that the relative gender gap is larger in eastern than in western Europe. These results show that tobacco plays an important role in generating international differences in the size of gender gap in mortality. The much discussed association between' the overall life expectancy and the gender gap in life expectancy is a numerical product of absolute death rates (differences in life expectancy are driven by differences in rates). The association of overall mortality with male/female mortality ratios is much weaker.