The idea of rural women as risk-averse food producers has been powerful and persistent and constitutes one of our most enduring generalizations. This contribution begins with some critical thoughts about the prevalent consensus on women and risk behaviour and goes on to discuss some counter examples of risk-taking women farmers in Zimbabwe and Zambia. It argues that risk behaviours of these kinds are strongly related to the character of marriage and forms of conjugality, and considers more broadly how insurance and dependence are gendered. There is a danger of overdrawing, and exclusively emphasizing, household and marriage as sites of gender subordination and thereby losing sight of the value to women of domestic groups and the existence of class-based solidarities and emotional investments, across gender, which are intertwined with gender subordinations. To recognize these (and other) positive aspects of institutions of kinship and marriage, without simultaneously endorsing subordination, requires a focus on change and women's agency within such institutions, and the happy thought that there is no such thing as the status quo. This study therefore considers myths made within different but overlapping contexts; first the idea of women as reliably risk averse (as well as disadvantaged in access to insurance) which holds sway in international development organizations and some gender analysis; and second, the myth of households as composed of entirely separate individuals with opposed gender interests, in which marriage is predominantly a contract legitimating the exploitation of women. Marriage works as a safety net for women in many contexts, as a form of insurance, but it may become an impediment to accumulation — a feature shared with other social security institutions.