The objective of this paper is to estimate the impact of medium- and long-term participation in microcredit programs. It utilizes a new, and large, panel dataset collected from treatment and control households from 1997 to 2005. The data enables us to identify continuing participants in the program as well as newcomers and leavers. We employ different estimation strategies including difference-in-difference-in-difference and propensity score matching methods to control for selection bias. The impact estimates indicate that the benefits from microcredit vary more than proportionately with the duration of participation in a program. Larger benefits are realized from longer-term participation, and the benefits continue to accrue beyond departure from the program. The findings indicate the need to observe longer periods of participation to provide a reliable basis for assessing the effectiveness of microcredit lending.