DescriptionOne of the most fascinating questions for evolutionary scientists is “How did humans arise as a new species?” In the last seventy years, two major schools of theory, allopatric speciation and sympatric speciation, have been developed and applied to explain the speciation process. Allopatric theory attributes the inducement of speciation to the establishment of geographic barriers that abruptly divide the ancestral population into two reproductively isolated groups, while sympatric theory emphasizes the role of divergent selection, leading to assortive mating and gradually diminishing gene flow. The two different scenarios should leave distinct footprints in the derivative genomes of the emerging species. Many mathematical methods have been developed to study human-chimpanzee speciation history by studying the genetic variation pattern in current human and chimpanzee populations. However, most methods either fail to incorporate sympatric speciation, or use datasets that don’t provide enough information about ancient divergence. In this study, we developed a new maximum likelihood method for analyzing genome data under the ‘isolation with migration’ model. Testing with simulated datasets demonstrates that this method is capable of generating accurate estimates regarding both current and ancient evolutionary histories. We applied this method to the whole-genome alignment of human, chimpanzee and orangutan. The estimated human-chimpanzee speciation time is 4.3 million years (Myr). This estimate is in agreement with several previous studies. A more important finding of our study is a weak but significant one-way gene flow from the chimpanzee to the human population (0.002 migrations per generation). Simulation studies confirm that this gene flow is not an artifact created by within-locus recombination or violation of other assumptions of our method. A further analysis finds that the gene flow from chimpanzees into humans and chimpanzees persisted for a limited period of time, subsequent to the initial separation. These results lead us to favor a speciation process for humans and chimpanzees that includes some limited genetic exchange.