A method of projection spectroscopy for N-dimensional NMR experiments with the following steps. Data recording through; a) selection of N-dimensional NMR experiments out of a group of N-dimensional experiments, selection of the dimensionalities (D.sub.i) of the projections and unconstrained selection of j sets of projection angles, with j.gtoreq.2; b) recording of discrete sets of j projections from the N-dimensional NMR experiments at the selected projection angles; c) peak picking and creating a peak list for each of the j projection spectra is characterized by d) automated identification of peaks in the projection spectra that arise from the same resonance in the N-dimensional spectrum (N.gtoreq.3) using vector algebra to exploit geometrical properties of projections in the N-dimensional space, and computation of a N-dimensional peak list using vector algebra to exploit geometrical properties of projections in the N-dimensional space. A reliable method of automated projection spectroscopy without restrictions on projection angles and dimensionality is thereby realized.