There is spirited debate between those who maintain that public assistance to the poor decreases poverty by raising their incomes (an income enhancement effect) and those who contend that welfare increases poverty by discouraging the poor from working (a work disincentive effect). Extant studies have been inconclusive because they have focused on the effect of welfare benefits on the poverty rate, but have not employed designs that allow researchers to sort out distinct income enhancement and work disincentive effects. We develop a model of poverty rates in the American states that permits estimation of these distinct effects based on state-level time-series data observed annually for the years 1960-90 and we find that welfare had both effects during our period of analysis. We also calculate the net impact on the poverty rate of an increase in welfare benefits (taking into account both income enhancement and work disincentives), and we conclude that it has varied across states and time. In general, however, the ability of marginal increases in welfare spending to reduce the poverty rate by enhancing incomes has declined since the 1970s.