This paper presents a Bayesian Markov chain Monte Carlo method for inferring species trees from multilocus data. The method coestimates multiple gene trees embedded in a shared species tree along with the effective population size of both extant and ancestral species. It uses multilocus data from multiple individuals per species to infer the species tree. The method is compared to existing methods such as BEST 2.2 and the supermatrix (concatenation) method. The results show that both BEST and the new method have much better estimation accuracy for species tree topology than concatenation, and the new method outperforms BEST in divergence time and population size estimation. The method is also tested on simulated data and real-world data from pocket gophers. The results show that the new method is more accurate in estimating species tree topology, divergence times, and population sizes compared to other methods. The method is implemented in the BEAST software package, which is a Bayesian phylogenetic inference tool. The method uses a multispecies coalescent model to account for incomplete lineage sorting and ancestral polymorphism. The method is able to handle multiindividual data and missing data, and provides a natural way to include these into phylogenetic analysis. The method is also able to estimate the total number of species and assign individuals to species. The paper discusses the limitations of the method, including the lack of modeling of recombination within a locus and the treatment of speciation events. The paper concludes that the multispecies coalescent represents a step toward the unification of molecular systematics and phylogeography. The method is able to provide accurate estimates of species tree topology, divergence times, and population sizes, and is able to handle multiindividual data and missing data. The method is implemented in the BEAST software package, which is a Bayesian phylogenetic inference tool. The method is able to handle multiindividual data and missing data, and provides a natural way to include these into phylogenetic analysis. The method is also able to estimate the total number of species and assign individuals to species. The paper discusses the limitations of the method, including the lack of modeling of recombination within a locus and the treatment of speciation events. The paper concludes that the multispecies coalescent represents a step toward the unification of molecular systematics and phylogeography.This paper presents a Bayesian Markov chain Monte Carlo method for inferring species trees from multilocus data. The method coestimates multiple gene trees embedded in a shared species tree along with the effective population size of both extant and ancestral species. It uses multilocus data from multiple individuals per species to infer the species tree. The method is compared to existing methods such as BEST 2.2 and the supermatrix (concatenation) method. The results show that both BEST and the new method have much better estimation accuracy for species tree topology than concatenation, and the new method outperforms BEST in divergence time and population size estimation. The method is also tested on simulated data and real-world data from pocket gophers. The results show that the new method is more accurate in estimating species tree topology, divergence times, and population sizes compared to other methods. The method is implemented in the BEAST software package, which is a Bayesian phylogenetic inference tool. The method uses a multispecies coalescent model to account for incomplete lineage sorting and ancestral polymorphism. The method is able to handle multiindividual data and missing data, and provides a natural way to include these into phylogenetic analysis. The method is also able to estimate the total number of species and assign individuals to species. The paper discusses the limitations of the method, including the lack of modeling of recombination within a locus and the treatment of speciation events. The paper concludes that the multispecies coalescent represents a step toward the unification of molecular systematics and phylogeography. The method is able to provide accurate estimates of species tree topology, divergence times, and population sizes, and is able to handle multiindividual data and missing data. The method is implemented in the BEAST software package, which is a Bayesian phylogenetic inference tool. The method is able to handle multiindividual data and missing data, and provides a natural way to include these into phylogenetic analysis. The method is also able to estimate the total number of species and assign individuals to species. The paper discusses the limitations of the method, including the lack of modeling of recombination within a locus and the treatment of speciation events. The paper concludes that the multispecies coalescent represents a step toward the unification of molecular systematics and phylogeography.