ABSTRACT To establish an improved ribosomal gene sequence database as part of the Ribosomal Differentiation of Microorganisms (RIDOM) project and to overcome the drawbacks of phenotypic identification systems and publicly accessible sequence databases, both strands of the 5′ end of the 16S ribosomal DNA (rDNA) of 81 type and reference strains comprising all validly described staphylococcal (sub)species were sequenced. Assuming a normal distribution for pairwise distances of all unique staphylococcal sequences and choosing a reporting criterion of ≥98.7% similarity for a “distinct species,” a statistical error probability of 1.0% was calculated. To evaluate this database, a 16S rDNA fragment (corresponding to Escherichia coli positions 54 to 510) of 55 clinical Staphylococcus isolates (including those of the small-colony variant phenotype) were sequenced and analyzed by the RIDOM approach. Of these isolates, 54 (98.2%) had a similarity score above the proposed threshold using RIDOM; 48 (87.3%) of the sequences gave a perfect match, whereas 83.6% were found by searching National Center for Biotechnology Information (NCBI) database entries. In contrast to RIDOM, which showed four ambiguities at the species level (mainly concerning Staphylococcus intermedius versus Staphylococcus delphini ), the NCBI database search yielded 18 taxon-related ambiguities and showed numerous matches exhibiting redundant or unspecified entries. Comparing molecular results with those of biochemical procedures, ID 32 Staph (bioMérieux, Marcy I'Etoile, France) and VITEK 2 (bioMérieux) failed to identify 13 (23.6%) and 19 (34.5%) isolates, respectively, due to incorrect identification and/or categorization below acceptable values. In contrast to phenotypic methods and the NCBI database, the novel high-quality RIDOM sequence database provides excellent identification of staphylococci, including rarely isolated species and phenotypic variants.