The functional properties of molecular systems are generally determined by the sum of many weak non-covalent interactions, and therefore methods for predicting the relative magnitudes of these interactions is fundamental to understanding the relationship between function and structure in chemistry, biology and materials science. This review focuses on the Surface Site Interaction Point (SSIP) approach which describes molecules as a set of points that capture the properties of all possible non-covalent interactions that the molecule might make with another molecule. The first half of the review focuses on the empirical non-covalent interaction parameters, α and β, and provides simple rules of thumb to estimate free energy changes for interactions between different types of functional group. These parameters have been used to have been used to establish a quantitative understanding of the role of solvent in solution phase equilibria, and to describe non-covalent interactions at the interface between macroscopic surfaces as well as in the solid state. The second half of the review focuses on a computational approach for obtaining SSIPs and applications in multi-component systems where many different interactions compete. Ab initio calculation of the Molecular Electrostatic Potential (MEP) surface is used to derive an SSIP description of a molecule, where each SSIP is assigned a value equivalent to the corresponding empirical parameter, α or β. By considering the free energies of all possible pairing interactions between all SSIPs in a molecular ensemble, it is possible to calculate the speciation of all intermolecular interactions and hence predict thermodynamic properties using the SSIMPLE algorithm. SSIPs have been used to describe both the solution phase and the solid state and provide accurate predictions of partition coefficients, solvent effects on association constants for formation of intermolecular complexes, and the probability of cocrystal formation. SSIPs represent a simple and intuitive tool for describing the relationship between chemical structure and non-covalent interactions with sufficient accuracy to understand and predict the properties of complex molecular ensembles without the need for computationally expensive simulations.