Admixture contributions are widely studied in human populations and several methods are available based on allele frequencies or, more recently, on molecular information. We developed the allelic variance method (AVM) based on the molecular information extracted from the microsatellites under the single stepwise mutation model (SSMM). AVM circumvents the allele methods limitation of a lineal model and focus on the θ(4N e μ) parameter properties. Compared to allele frequency methods (AFMs), AVM has more accurate estimations and no systemic bias under different simulated scenarios. When it is applied to human data, the results of the analysis of three populations from Latin America (Argentina, Brazil, and Colombia) indicate a correct inference of a trihybrid model, confirmed by other three different methods. However, the estimates from AVM seem to discriminate better the admixture demographic histories of the Brazilian and Colombian populations.