In this paper we introduce some theoretical and methodological refinements to account for effects of duration dependence in forced migration flows that have not been noticed in previous research. The engine of previous theoretical arguments on forced migration has been a decision theoretic model in which the potential migrants estimate the threat to their security and then weigh this threat against costs and benefits of leaving. Moreover, in previous arguments and empirical tests the costs and benefits for relocating have been treated as being the same for all people in a country. We relax this assumption, allowing the costs and benefits of relocating to be different for different individuals. This implies that some people more readily relocate than others. Over time this generates a selection effect in the population that remains behind, such that the remaining population will become increasingly unwilling or unable to relocate. The implications of our theoretical refinement are borne out empirically. Contrary to previous research we find that the accumulated stock of forced migrants decrease rather than increase the probability of new migration. Furthermore, we find that forced migrant flows abate rather than soar over time.