The basic local alignment search tool (BLAST) is one of the most popular sequence alignment tools available. Sequence alignment is used to extract similar parts of an input protein (or DNA) sequence from protein (or DNA) databases, in order to investigate biological evolution and genomic genealogy. It is a very important and difficult task in bioinformatics. Even though BLAST is an efficient sequence alignment algorithm, it cannot cope with the rapid growth of databases. Initially, in a preprocessing step, BLAST creates query words and a neighborhood word list. Next, it performs three processing steps: a) seed search, b) ungapped extension, and c) gapped extension. In this paper, we propose a hardware accelerator to speed up all processing steps of BLAST, including preprocessing, related works speed up only a subset of them. Each processing step is realized by a hardware module. Thus, we can easily speed up each processing step by duplicating the corresponding module. By implementing the proposed accelerator in a field programmable gate array (FPGA), we demonstrate that a high-performance accelerator can be realized with reasonable hardware cost.