The estimation of population parameters from genetic data can help reveal p
ast migration patterns or past population sizes. The transformation of raw
genetic data into population parameters requires a model which should refle
ct the true relationships between subpopulations. Often the models are over
ly simplified and do not allow, for example, for differences in population
sizes and migration rates. I stress here the point that it is important to
consider possible asymmetries in migration rates and differences in populat
ion sizes. Very recently, several estimators based on the direct use of all
ele frequencies and based on coalescence theory have been developed. All th
ese outperform migration rate estimators based on F-ST.