It has been demonstrated that irrationality reduces the efficiency of individuals’ allocations, as measured by their “true” or rational preferences. There is also evidence that poverty increases irrationality of different sorts. As a result, the net benefit to society of a cash transfer from taxpayers to welfare recipients may not be zero. The fact that the transfer will be allocated less efficiently by the recipients than by the taxpayers will reduce the value of the transfer, while if the transfer increases recipients’ rationality, it will increase the efficiency of the allocation of their pretransfer budgets, thus increasing the value of the transfer. The net effect on society will be positive or negative, depending in large part on the degree to which the transfer increases rationality. I model these effects in the context of present-biased preferences and explore the effect of irrationality, income, and the size of transfer on the value of transfers. I conclude that under a plausible range of conditions, transfers can generate a substantial positive net benefit. I also model the choices of a fully rational paternalist and find little support for paternalistic in-kind transfers.