The distribution of vascular plants in grid-cells and its relationship to the environmental correlates (driving factors) were studied using numerical methods (divisive classification and ordination). The first level of division in the classification distinguished forest and non-forest groups of grid-cells, and the second level four groups (containing predominantly species of base-rich forests at high altitudes, species of acidophilous mountain forests and small mountain grasslands, ruderal and meadows species at low altitudes, and species of thermophilous and basiphilous fringes and abandoned meadows). Within the study area, geographically consistent areas were delimited by correlating the groups, indicated by the divisive classification, with altitude and forest cover. Most differences in the Ellenberg indicator values for species in these groups for light, temperature, continentality, soil reaction and soil moisture were statistically significant. A number of variables were effective predictors (e.g. potential direct solar irradiation), physical geography (altitude, slope), land-cover (forest cover, area of urban zones) and geological bedrock were the key determinants of the species composition in the study area. However, even the most spatially correlated (according to Moran’s I measure) were the naturally contiguous variables such as topographical features (altitude, slope and aspect). Generally, the grid-cells at low altitudes contained more species due to the co-occurrence of man-made habitats with fragments of semi-natural habitats. A relatively large percentage of the variation (15.8%) was accounted for by the spatial structure of the data, the environmental factors explained 18.9%, but 65.3% of the floristic variance remained unexplained. The most spatially autocorrelated variables were also the most correlated with regard to species composition. However, the relatively high autocorrelation in the species data and their derivates had comparable or lower effect on species composition than the most autocorrelated environmental factors. The results were compared with those of other European studies, and possible bias due to the different ways of collecting and analysing data, and effect of different scales discussed.