We study the effects of unanticipated changes in the intra-household division of parental leave on family stability by exploiting two parental leave reforms in Sweden. Using a fuzzy regression discontinuity design, we find that an unexpected change in the mother’s share of parental leave increases the probability of separation among couples that were married or cohabiting at the time of the reforms. Our results also suggest a lower likelihood of cohabiting couples to upgrade to marriage. Furthermore, we find some evidence that cohabiting couples react more strongly than married couples to changes in the division of parental leave, which could be driven by different costs of separation.