Multivariate statistical methods were used to analyse the vitality of Scots pines (Pinus sylvestris L.) by means of various biochemical, physiological, and nutritional characteristics, irrespective of tree age and without site-specific information. The vitality model was developed in three steps. First, artificial neural networks were used to select a minimal set of biomarkers as input variables with respect to the output variable circular surface increment (regression problem). In the second step, vitality states were classified by using the selected biomarkers (cluster analysis). Finally, discriminant analysis was applied to assign Scots pines to one of four classified vitality states. Sulfate sulfur (SO42--S), non-protein-nitrogen (NPN), arginine (Arg), and chlorophylla (Chla) proved to be the best dynamic input variables to reliably determine the vitality of Scots pines. As the results of regression problem solutions showed, the optimized neural network found growth responses to the driving variables that are valid for both young and mature pine stands. However, the sensitivity analysis of the neural network also indicated that of the four variables, sulfate is the least sensitive. Nevertheless, the sulfate response of the network can be successfully used to analyze the specific effects of multiple exposure on the vitality of Scots pines. To summarize final vitality model enables the vitality of Scots pines to be evaluated without reference to tree age or knowing the specific forest conditions.