We present a model in which there is potentially conflict within households about resource allocations. These conflicts are unlikely to be perfectly bargained out and hence there is some residual inefficiency associated with conflict which ought to increase with household size. We also show that individuals who contribute more than their fair share to household resources are likely to leave larger households and households riven with more conflicts sooner than smaller or more harmonious ones. These implications are testable in principle and we provide some evidence from South Africa which is consistent with the theory.