The quantitative structure activity relationship (QSAR) methodology has been developed and extensively used to predict unknown environmental data for compounds that have not been experimentally studied yet. QSAR is based on a large series of descriptors: such as the number of atoms, the number of bonds… (descriptive), or based on the 2D structure of the molecule (connectivity indices…) or on its 3D structure (dipole moment, polarizability…). Among them, quantum-based 3D descriptors appear as promising tools to predict macroscopic environmental properties. For a set of 104 pharmaceuticals and personal care products, four quantum-based 3D descriptors (electric dipole moment, polarizability, HOMO energy and ionization potential) were calculated using different computational chemistry strategies involving a conformational search followed by local quenches within three different frameworks: density functional theory (DFT), semi-empirical Austin Model 1 (AM1) approach, and density functional based tight binding (DFTB). Comparing the results obtained using each framework highlights the necessity of a comprehensive conformational search and the use of an accurate potential for the local quenches. Using the combination of a global exploration through molecular dynamics with local quenches at B3LYP/6-31G* (DFT) allows the calculation of accurate and tractable quantum-based 3D descriptors.