Diagnostic (character, differential) species are species with a distinct concentration of occurrence or abundance in a particular vegetation unit. These species are valuable for identification of earlier described vegetation units in field survey or monitoring projects. The standard vegetation classification of the Czech Republic (MORAVEC et al. 1995) lists groups of diagnostic species for phytosociological alliances and suballiances; however, these lists are based on expert knowledge and involve many inconsistencies. In the present study, we used a geographically stratified data set of 21,794 phytosociological relevés selected from the Czech National Phytosociological Database to define diagnostic species for all phytosociological classes, alliances and suballiances of the standard national vegetation classification in a consistent way, with statistical methods. Diagnostic species were defined from this data set as those exceeding the selected threshold of statistical fidelity to vegetation units. Fidelity was calculated as the Phi coeffient, which measures statistical association between vegetation units and species occurrences. In addition, we determined constant species, i.e. those with frequent occurrence, and dominant species, i.e. those frequently attaining a high cover in particular vegetation units. Having determined diagnostic species, we were able to evaluate the quality of delimitation of classes, alliances and suballiances of the standard national classification. We used two criteria of quality, sharpness and uniqueness, both based on diagnostic species. Sharpness of a vegetation unit was defined as a proportion or quality of its diagnostic species relative to the average species richness of its relevés. The sharpest vegetation units mainly included the major types of natural and semi-natural vegetation of Central Europe. By contrast, low sharpness was detected for those vegetation units which are either fragmentarily developed in the Czech Republic (though well-developed abroad) or poorly delimited in their entire geographic range. Uniqueness of a vegetation unit corresponded to the relative number of its diagnostic species which were not simultaneously diagnostic for other vegetation units. Low uniqueness indicated those vegetation units which could be possibly merged with some other units. The analysis of sharpness and uniqueness identified well defined and poorly defined vegetation units of the standard national vegetation classification. In such a way, it suggested possibilities of alternative delimitations of some vegetation units in a future revision of this classification.