Several lines of evidence argue in favour of a recent and unique origin of modern humans in sub-Saharan Africa, but no attempt has really been made at quantifying the likelihood of this model, relative to alternative hypotheses of human evolution. In this paper, we investigate the possibility of using multi-locus genetic data to correctly infer the geographic origin of modern humans, and to distinguish between a unique origin (UO) and a multiregional evolution (ME) model. We use an approach based on realistic simulations of the genetic diversity expected after an expansion process of modern humans into the Old World from different possible areas, and with different environmental scenarios, under both UO and ME models. We find that UO and ME models should produce distinctive patterns of genetic diversity in observed data. Moreover, the geographic origin of an expansion is recoverable under the UO model provided that a large number of independent markers are used, and that precise information on past demography and potential places of origin(s) are available. We also find that the successful recovery of past scenarios is related to the degree with which environmental heterogeneity impacts on past demography and migration, suggesting that use of a realistic representation of past environment is important to make correct inferences. Finally, the application of our simulation framework to the problem of the origin of mankind clearly rejects multiregional evolution scenarios, and points toward a unique and African origin of modern humans.