Newly available DNA sequencing technologies can generate billions of DNA sequences in a single machine run, making it feasible to obtain an individual’s entire genome sequences very quickly and at a very affordable cost. These personal genome sequences can be used to identify genetic variations associated with variable traits, but they must first be aligned to a reference genome sequence using computer algorithms that make use of approximate string matching methods. The process of aligning a very large number of short sequence reads is computationally expensive and has led to the development of many sequence mapping programs. While new software, such as BWA, SOAP2, and Bowtie, efficiently align large numbers of short DNA sequences, we proposed that parallel computing would provide additional speedup in data processing to accommodate the rapidly increasing sequencing throughput. For this purpose, we developed pBWA, an efficient parallel version of BWA, based on the Open MPI library. pBWA retains the multi-threading capability provided by BWA while adding efficient parallelization for its core alignment functions. We have shown that pBWA’s wall-time speedup is bounded only by the size of the parallel system. pBWA can run in both parallel and multi threaded environments simultaneously, allowing it to be tailored to run on parallel systems of all configurations and sizes. By taking the advantage of high performance computing cluster, the use of pBWA can cut down the computing wall-time for the reference alignment step from weeks to hours for extremely large DNA sequence data. We expect that the availability of pBWA should facilitate the analysis of large-scale genome sequencing data generated by the new generations of sequencing technologies. The source code and detailed user manual of pBWA are freely available at http://sourceforge.net/projects/pbwa.