Many ecological data are compositional and different quantitative techniqueshave been used to analyze such data, albeit some of them being methodologically wrong. The aim of this contribution is to apply the compositional data approach to forestry data and demonstrate the strengths of this method for percentage or relative data with infrequent zero values. Basal areas of three dominant tree species (Abies alba, Picea abies, Fagus sylvatica) in 119 forest compartments in some of the Omphalodo-Fagetum forests in Slovenia in 1954 and 2004 were used to investigate the dynamics of forest species composition over a 50-year period. For the investigated period some additional data about geomorphology and harvesting rates within the compartments were used as explanatory variables of compositional change. The species composition of each forest compartment was subjected to several methods within a compositional analysis framework: descriptive, ternary diagram-based graphical presentations, significance of compositional differences between management classes, significance of perturbation differences (the indicator of forest compositional change) and relation of the compositional change with the explanatory variables by means of compositional linear model. Results indicated that the silver fir was the dominant species in both years, but a clear reduction in silver fir proportion was observed after 50 years. The perturbation differences indicated comparatively large relative increase in the proportion of Norway spruce between 1954 and 2004. Subsequently, the perturbation differences were subjected to isometric log-transformation (ilr) and two derived ilr coordinates were further used as dependent variables in the multivariate linear model. The initial stand structure correlated well with the perturbation differences. These were also significantly correlations with salvage cutting, a consequence of silver fir decline in the 1954-2004 period. This study demonstrated that the compositional data approach can be successfully used to study forest dynamics yielding some insights into data which are not possible or even not valid using some alternative methods.