In this study, we established an in-house database of yeast internal transcribed spacer (ITS) sequences. This database includes medically important as well as colonizing yeasts that frequently occur in the diagnostic laboratory. In a prospective study, we compared molecular identification with phenotypic identification by using the ID32C system (bioMérieux) for yeast strains that could not be identified by a combination of CHROMagar Candida and morphology on rice agar. In total, 113 yeast strains were included in the study. By sequence analysis, 98% of all strains were identified correctly to the species level. With the ID32C, 87% of all strains were identified correctly to the species or genus level, 7% of the isolates could not be identified, and 6% of the isolates were misidentified, most of them as Candida rugosa or Candida utilis. For a diagnostic algorithm, we suggest a three-step procedure which integrates morphological criteria, biochemical investigation, and sequence analysis of the ITS region.