Michael Bamshad and colleagues report sequencing of 15,585 protein-coding genes to an average depth of 11× in 2,440 individuals of European and African ancestry (Science, published online 17 May 2012; doi:10.1126/science.1219240). They identify over 500,000 single-nucleotide variants (SNVs), the majority of which are rare, novel and population specific. The overall site frequency spectra (SFS) were highly skewed, showing a large excess of rare variants compared to the number expected with the standard neutral model. The authors fit a modified out-of-Africa demographic model that provides support for a recent and rapid acceleration in population growth. In characterizing functional variation using seven different computational methods, the authors found that ~47% of SNVs were predicted to be deleterious by at least one of these methods, with limited overlap in predictions between methods. The authors estimated a mean of 13,595 SNVs and 549 novel SNVs per individual, with both of these numbers higher in individuals of African ancestry relative to those of European ancestry. Further, they estimated an average of 318–580 predicted functional protein-coding SNVs per individual, depending on the definition of functional variant, and this estimate was higher in individuals of African ancestry than in those of European descent.