This paper presents a computer-based method to estimate optimal migration routes of early human population groups by a combination of ecological niche analysis and least-cost path analysis. In the proposed method, niche probability is predicted by MaxEnt, an ecological niche model based on the maximum entropy theory. Location of known archaeological sites and environmental factors derived from palaeoterrain and palaeoclimate models, are input to the model to calculate the niche probability at each spatial pixel and weights of the environmental factors. The inverse of probability score is then used as an index of relative dispersal rate to accumulate the travel cost from a given origin. Based on this cumulative cost surface, least-cost paths from the origin to given destinations are visualised. This method was applied to the Initial Upper Palaeolithic population group (probably of modern humans) in Eurasia. The model identified three migration routes from the Levant to (1) Central Europe via Anatolia and Eastern Europe, (2) the Russian steppe via Caucasus Mountains, and (3) the Altai region via the southern coastal Iran and Afghanistan.