We examine in depth the effect of differences in the smoking adoption patterns of men and women on the mortality gender gap in Netherlands, employing a historical perspective. Using an indirect estimation technique based on observed lung cancer mortality from 1931 to 2012, we estimated lifetime smoking prevalence and smoking-attributable mortality. We decomposed the sex difference in life expectancy at birth into smoking-related and nonsmoking-related overall and cause-specific mortality. The smoking epidemic in Netherlands, which started among men born around 1850 and among women from birth cohort 1900 onwards, contributed substantially to the increasing sex difference in life expectancy at birth from 1931 (1.3 years) to 1982 (6.7 years), the subsequent decline to 3.7 years in 2012, and the high excess mortality among Dutch men born between 1895 and 1910. Smoking-related cancer mortality contributed most to the increase in the sex difference, whereas smoking-related cardiovascular disease mortality was mainly responsible for the decline from 1983 onwards. Examining nonsmoking-related (cause-specific) mortality shed new light on the mortality gender gap and revealed the important role of smoking-related cancers, the continuation of excess mortality among women aged 40–50, and a smaller role of biological factors in the sex difference than was previously estimated.