Extensive laboratory evidence of declining time discount rates motivated models of (quasi) hyperbolic discounting. These models are now being widely used to describe the self-control problems that cause individuals to seek costly commitments to future choices. This paper reports the results of a field experiment in rural Malawi designed to evaluate the importance of time discounting for self-control and demands for commitment. Adapting the methods of Andreoni and Sprenger (2010), the experiment uses real choices over money to elicit two time discount rates (near and far-term) and intertemporal elasticities of substitution for approximately 2,200 respondents. The stakes of the intertemporal choices are high, representing a total of approximately one month’s rent per individual. Having made their committed intertemporal choice, a randomly selected subset of respondents is approached at randomly selected points in an interval of 10 days prior to the arrival of the first instalment of money. This subset is asked if they would like to revise their initial allocation and, in this way, potentially exhibit time-inconsistency. The paper then relates revision choices to estimates of individual preference parameters as well as survey evidence about the arrival of shocks, demands from friends and relatives, intra-household conflict over preferences and abilities at financial calculation. A key question is whether revising decisions depend on how close they are in time to the arrival of the first instalment of money. In this way, the paper evaluates the relative importance of time discounting in determining time preference and time-inconsistency.