The current theory and methods of phytosociological vegetation classification are divided into two major types: imperfectly formalized approaches, which build classifications without explicit description of the classification process, and formalized approaches, which aim at precise definitions of classification criteria and algorithms, thus yielding repeatable classifications. These two approaches are not antagonistic, as each of them is better applicable in different situations. As a rule, the imperfectly formalized approach is invaluable in fine-scale classifications at the landscape level, whereas the formalized approach is superior for large-scale vegetation surveys. In this paper, application of the formalized methods of vegetation classification in phytosociology is reviewed. First, classification criteria used in vegetation classification are evaluated. Second, possibilities for formalizing the phytosociological field sampling procedure are discussed, including the sample plot choice, spatial arrangement and size. Third, formalized approaches to data analysis are reviewed, including concepts of character species, sociological species groups, fidelity, hierarchy of classification units, deductive classification method, numerical classification, and nomenclature rules. Finally, recent developments of formalized classification are summarized with respect to large-scale vegetation surveys, phytosociological databases, analysis of large datasets and expert systems.