This paper applies Carter and Barrett’s theory of assets poverty traps to a unique longitudinal survey from rural Bangladesh. Non-parametric and parametric methods are used to examine the shape of the dynamic asset frontier, the number of equilibria, and whether land and nonland assets stock converge to such equilibria.  We find evidence for concavity of the dynamic asset frontier but no evidence for multiple equilibria in the case of both land and non-land assets. It is hypothesised that the existence of well-functioning markets for labour and capital and the absence of discrete differences in livelihood strategies in rural Bangladesh, and Asia more generally, help to explain the contrast between our results and those for several African countries.