We study the malleability of gender-specific preferences for work by exploiting the German division and reunification as a natural experiment. We test whether the two political systems have shaped gender gaps in preferences differentially, based on German-General-Social-Survey data from 1991, 1998 and 2012, an extensive set of register data and historical data from the 19th and early 20th century. Our analyses reveal a substantial East-West difference in the gender gap directly after reunification and no convergence thereafter. A cohort analysis illuminates the mechanism, as the effect is driven by cohorts who grew up during separation, and suggests that institutions, not cultural legacy, are the decisive component.