The authors try to determine whether children sent to work in rural Bangladesh are caught in a poverty trap, with the extra income to poor families from child labor coming at the expense of the children's longer-term prospects of escaping poverty through education. The poverty trap argument depends on children's work being substitutable for schooling. Casual observations and the descriptive statistics available from surveys seem to offer little support for the argument. To explore the question more deeply, the authors use a targeted school stipend to identify how much child labor substitutes for schooling. They find that Bangladesh's Food-for-Education program is a strong incentive for school attendance. A stipend with a value considerable less than the mean child wage was enough to ensure nearly full school attendance among participants. The enrollment also reduced the incidence of child labor, an effect accounted for only a small proportion of the increase in school enrollment. The reduction in the incidence of child labor among boys (girls) represents about one-quarter (one-eighth) of the increase in their school enrollment rate. Parents are clearly substituting other uses of their children's time to secure income gain from access to the program, with modest impact on earnings from their children's work. The authors'tests were limited. Work may well displace time for doing homework or attending after-school tutorials, for example. The authors were unable to identify such effects from the data available. There may also be other welfare losses to children from work (such as exposure to an unsafe working environment) as well as welfare gains (such as skills learned from working that enhance returns to schooling). But their results do lead them to question the seemingly common view that child labor is a major factor perpetuating poverty in Bangladesh by keeping children from poor families our of school.