The very poor in developing countries often make intertemporal choices that seem at odds with their individual self-interest. There are many possible reasons why. This paper investigates several of these reasons with a lab-in-the-field experiment in rural Malawi involving large stakes. It makes two contributions. First, it constructs a new dependent variable: revisions of prior choices regarding the allocation of future income. This allows us to directly examine intertemporal choice revision and its determinants. In particular, this dependent variable permits a novel test for the existence of self-control problems. It turns out revisions of money allocations toward the present are positively associated with measures of present-bias from an earlier baseline survey, as well as the (randomly assigned) closeness in time to the first possible date of money disbursement. Second, the paper investigates other potential determinants of revision, aside from self-control problems. It finds little evidence that revisions of money allocations toward the present are associated with spousal preferences for such revision, household shocks or the financial sophistication of respondents.