Family break-up is common in families experiencing homelessness. This paper examines the extent of separations of children from parents and of partners from each other and whether housing and service interventions reduced separations and their precursors among 1857 families across 12 sites who participated in the Family Options Study. Families in shelters were randomized to offers of one of three interventions: permanent housing subsidies that reduce expenditures for rent to 30% of families’ income, temporary rapid re-housing subsidies with some services directed at housing and employment, and transitional housing in supervised facilities with extensive psychosocial services. Each group was compared to usual care families who were eligible for that intervention but received no special offer. Twenty months later, permanent housing subsidies almost halved rates of child separation and more than halved rates of foster care placements; the other interventions did not affect separations significantly. Predictors of separation were primarily homelessness and drug abuse (all comparisons), and alcohol dependence (one comparison). Although housing subsidies reduced homelessness, alcohol dependence, intimate partner violence, and economic stressors, the last three variables had no association with child separations in the subsidy comparison; thus subsidies had indirect effects via reductions in homelessness. No intervention reduced partner separations.