Aims: The forest vegetation of Turkey has been studied extensively using the Braun-Blanquet approach, but mostly in rather small areas which were not sufficient to provide an integrative phytosociological interpretation. In this study, we formed a regional forest vegetation database and aimed to construct a hierarchical syntaxonomic system using numerical methods. Study area: The Western Euxine province of the Euro-Siberian phytogeographic region of Turkey, extending from the province of Sinop in the east to the Istranca Mountains in the west. Methods: Vegetation data consist of 1526 Nieves and 872 species. The dataset was split into coniferous forests and deciduous-mixed forests. For the delimitation of these categories, all releves where the total cover of deciduous trees reached at least 1/3 of the total cover of tree layer were defined as deciduous-mixed forests, while releves with a total cover of deciduous trees less than 1/3 were defined as coniferous forests. Classification of these forests was carried out using TWINSPAN. Diagnostic species of vegetation units were defined using both the phi coefficient and constancy ratio. For tree species, the ratio of average cover was taken instead of constancy ratio. The vegetation units were further evaluated using DCA ordination and comparison of environmental variables. Results: The classification of deciduous-mixed forests resulted in 21 units differentiated by edaphic, climatic or phytogeographic factors. These communities were assigned to three classes (Quercetea pubescentis, Carpino-Fagetea, Alno-Populetea), four orders and five alliances. The oak-hornbeam forests of Turkey were described as a new alliance Trachystemono orientalis-Carpinion betuli since there was no valid alliance name available. For coniferous forests, eight communities were determined which are classifed in three classes (Quercetea ilicis, Erico-Pinetea, Vaccinio-Piceetea), three orders and three alliances. Mesic fir forests were assigned to a new alliance Lonicero caucasicae-Piceion orientalis within a new order Abieti nordmannianae-Piceetalia orientalis. Conclusions: In separating coniferous and deciduous-mixed forests prior to the numerical classification, we were able to achieve a classification in good accordance with the current syntaxonomic system used in Europe. The proposed system of higher syntaxa, which is mainly based on the dominant tree species, will provide a more stable and objective framework for future studies on biodiversity, nature conservation and sustainable forest management. Moreover, we provide an overview of the associations corresponding to each vegetation unit.