Population genomic analyses have demonstrated power to address major questions in evolutionary and molecular microbiology. Collecting populations of genomes is hindered in many microbial species by the absence of a cost effective and practical method to collect ample quantities of sufficiently pure genomic DNA for next-generation sequencing. Here we present a simple method to amplify genomes of a target microbial species present in a complex, natural sample. The selective whole genome amplification (SWGA) technique amplifies target genomes using nucleotide sequence motifs that are common in the target microbe genome, but rare in the background genomes, to prime the highly processive phi29 polymerase. SWGA thus selectively amplifies the target genome from samples in which it originally represented a minor fraction of the total DNA. The post-SWGA samples are enriched in target genomic DNA, which are ideal for population resequencing. We demonstrate the efficacy of SWGA using both laboratoryprepared mixtures of cultured microbes as well as a natural host-microbe association. Targeted amplification of Borrelia burgdorferi mixed with Escherichia coli at genome ratios of 1:2000 resulted in .10 5 -fold amplification of the target genomes with ,6.7-fold amplification of the background. SWGA-treated genomic extracts from Wolbachia pipientis-infected Drosophila melanogaster resulted in up to 70% of high-throughput resequencing reads mapping to the W. pipientis genome. By contrast, 2-9% of sequencing reads were derived from W. pipientis without prior amplification. The SWGA technique results in high sequencing coverage at a fraction of the sequencing effort, thus allowing population genomic studies at affordable costs. C LASSICAL population genetics, coupled with advances in coalescent modeling, has been foundational to studies of the evolutionary histories and ecological forces that shape natural populations (Rosenberg and Nordborg 2002;Hume et al. 2003;Wakeley 2004). However, detecting fine scale processes using population genetics and coalescent analyses is limited by the amount of available sequence data per sample. Datasets with substantially greater genetic information per sample, such as genomic data from population-level sampling, would be optimal to study biological processes at all relevant scales. The promise of population genomics for many microbial species is tempered, however, by the difficulty of isolating and preparing microbial genomes for nextgeneration sequencing. Currently, sequencing microbial genomes requires laboratory culture to isolate them from other organisms with which they are naturally associated to obtain the appropriate samples for sequencing-sufficient numbers of the target genome with limited contaminating DNA (Mardis 2008).Methodological issues in obtaining populations of genomes from microbial species occur both because the target microbial genomes often constitutes only a miniscule fraction of the DNA in complex, field-derived samples and because many important microbial specie...