There are several formats to describe PDB structural data and I introduce them in this article. Each format file is available from the PDBj web site (Figure 1). The most historical format is PDB format (Figure 2). It is 80 columns fixed format and the included information is partial. Moreover it includes some free description sections and it is not suitable for computational processing. As it will be phased out in 2016, it is recommended to use the current canonical format PDBx/mmCIF (Figure 3). It is defined by the extension of STAR format, which can include flexible length of lines. wwPDB also provides XML translated format PDBML (Figure 4). One can use general XML purser for it, but the file size is larger than PDBx/mmCIF. In the PDBx/mmCIF and PDBML, two types of ids for items such as atoms and residues are defined. One is author-defined id, which is not always sequential and may include non-numerical strings. The other is defined by wwPDB, which follows the systematic rule.