An analytical framework is suggested for the economic evaluation of policies to improve the management of hospital infection. Consideration of the costs and benefits to be expected from improved policies implies the existence of an optimal infection rate which is higher than the minimum attainable. It follows that hospitals can and probably do spend too much on infection control in, at least, some areas. This optimal approach is not operational at present but its data requirements might be thought of as an agenda for future research. In the meantime, progress in infection control economics depends on a piecemeal approach. An example is given in the estimation of the cost savings attributable to an improved antibiotic prescribing policy at the Royal Liverpool Hospital.