Agricultural development is crucial in developing countries, and particularly in the poorest countries where it accounts for large shares of employment and income and whose poverty is due simply to having a large share of the workforce in low-productivity agriculture. Raising productivity in agriculture is critically important for development, as is smoothly moving workers out of agriculture into more productive employment in other sectors. Raising agricultural productivity helps both to raise incomes and to reduce poverty-both by raising the incomes of poor people working in agriculture and by lowering the prices of foods that make up a disproportionately large share of the expenditures of poor people. In small and open economies, the in-crease in profitability of agriculture following improvements in productivity might tend to retain or even attract workers into agriculture. By contrast, at a global level, or at national level when policy focusses on self-sufficiency, improvements in agricultural productivity will free up labor for employment in other sectors. Incomes are generally much higher in non-agricultural work in developing countries-more than double those in agriculture after careful adjustment for key differences. This raises the possibility of a double dividend from structural transformation as workers move into higher-productivity activities. A key question for development policy is whether it is enough to simply evaluate the gains from higher productivity within agriculture, or whether potential benefits from structural change be included as well. This paper examines the arguments on this question. It concludes that these dividends may be substantial-but whether they are or not depends on the source of the initial differences in productivity and on the direction of movement when agricultural productivity rises. If it results from policy barriers such as restrictions on the transfer of farmland or requirements for residence permits in urban areas, there are likely to be substantial welfare gains when labor moves out of agriculture. They may also be substantial if urban wages are artificially high and attract substantial numbers of job-waiters into unemployment. However, these gains may be illusory if the income gaps arise primarily from differences in skills or from reluctance to move created by asset fixity.