It has been suggested that altering the pace of reproduction would improve the health of women and children. For formulating intervention policies, it is important to know whether on its own such a strategy is likely to lead to risk reduction. This paper analyses mortality risk in sibships to explore the relationship between family formation factors and other household characteristics that identify women whose families are at higher risk. The analysis allows for the fact that reproductive behaviour may be modified by the family's prior experience of child death, using simultaneous equations methods to purge the model of the ‘feedback’ effects of death on the endogenous variable, childbearing pace. The strong relationship between reproductive pace and average risk in a family appears to be due to the association of both with other differences between households. Other aspects of family formation patterns are good indicators of which families are likely to experience excess risks to their children. These factors are associated with maternal education, but measure characteristics of the family or mother that educational attainment does not fully capture. They indicate that high-risk mothers are likely to have less control over many aspects of their lives. The pace of family building does not lead to excess average family risk, but may result, at least in part, from the concentration of risk in families with other characteristic patterns of family formation and few resources. The paper argues for a broader conception of household influences on child health and the health-related behaviour of parents.