Sustainable forest management requires information on a number of soil properties. Therefore fast methods of soil analysis are needed. The objective of this study was to test the ability of near infrared (NIR) spectroscopy to predict the total and exchangeable Na, K, Ca, Mg, Mn, Fe and Al, the cation exchange capacity (CEC), the base saturation (BS) and the total contents of Zn, Cu, Cd and Pb in geologically heterogeneous forest soils. The samples (n = 100) were collected from five sites covered by beech (Fagus sylvatica L.) forest stands and from four depths. The soils were analysed for total contents of Na, K, Ca, Mg, Mn, Fe, Al, Zn, Cu, Cd and Pb after pressure digestion in HNO 3 and for contents of exchangeable Na, K, Ca, Mg, Mn, Fe and Al in NH 4 Cl. The spectra were recorded between 400 and 2500 nm at 2 nm intervals. Principal components analysis revealed significant spectral differences among the samples from different sites. The global NIR models predicted well the mean values of the total contents of all cations, Zn, Pb, the exchangeable K, Mg, Fe and BS in upper (0-10 cm) and lower (10-20 cm) layers of the studied soils: the correlation coefficients (r 2) of the linear regression (measured against predicted values) varied between 0.90 and 1.00 and the regression coefficients (a) ranged from 0.94 to 1.07. The contents of Cu, CEC and the exchangeable Na, Ca and Al were predicted satisfactorily (r 2 = 0.87-0.98, a = 0.86-1.14). The global models overestimated the values of total Ca and exchangeable Ca, Mn and Fe in the lower parts of their ranges resulting in biased estimations of the means at some of the considered sites. Splitting the sample population into spectrally similar groups enabled the development of local calibrations, which improved the prediction accuracy (lower standard errors of prediction) for most of the analysed constituents and removed the bias in the estimations of exchangeable of total Ca and Ca, Mn and Fe. The obtained results indicated the usefulness of NIR spectroscopy for determination of a number of soil constituents in geologically heterogeneous forest soils.