Income transfers from social programs are often not gender neutral and should, according to the vast literature on intra-household decision making and allocation, affect the distribution of bargaining power within the household. This result, however, was by and large established under the assumption of marriage stability. If this assumption does not hold, then the positive response of bargaining power to income found in the empirical research may be the artefact of sample selection. One may postulate, however, that when restricted to certain groups in the population, such as seniors, the assumption may hold since their probability of divorce is close to zero. In this paper we prove that the assumption is wrong, even when applied to seniors. We use a non-contributory pension reform in Argentina, that resulted in an unexpected and substantial increase in permanent income for around 1.8 million women, to study its effects on outcomes related to both marital stability and women’s bargaining power within the household. We find that the reform increased the probability of divorce/separation among senior highly educated women but had no impact on the low-educated. Instead, the latter gained considerable bargaining power within the household by decreasing the probability of being the only one in charge of household chores in tandem with an increase in their husbands’ participation in these chores.Centro de Estudios Distributivos, Laborales y Sociales (CEDLAS)