This paper introduces a new approach to probing intermolecular interactions based on a framework of two-dimensional (2D) synchronous spectroscopy. Mathematical analysis performed on 2D synchronous spectra using variable concentration as an external perturbation shows that the cross-peaks are composed of two parts. The first part reflects intermolecular interactions that manifest in the form of deviation from the Beer-Lambert law. The second part is related simply to the concentration variations of the solutes and is responsible for the generation of interfering cross-peaks not related to the intermolecular interactions in the system. It is the second part that prevents the reliable identification of intermolecular interactions. We propose a way of selecting the concentrations of solutes so that the resultant dynamic concentration vectors of different solutes become orthogonal to one another. Therefore, the contribution of the second part to the cross-peaks can be effectively removed by the dot product of orthogonal vectors. Our new approach has been tested on a simulated chemical system and a real chemical system. The results demonstrate that interfering cross-peaks can be successfully removed from a 2D synchronous spectrum so that the cross-peaks can be used as a reliable tool to characterize or probe intermolecular interactions.