A new kind of solvent descriptor obtained from quantum chemical calculations is introduced. Group contributions to each solvent descriptor are regressed for 71 UNIFAC groups. A reaction kinetic model is built by correlating a set of experimentally determined reaction rate constants in various solvents with the corresponding theoretical solvent descriptors. Based on the kinetic model and the developed group contribution method, a computer‐aided molecular design problem is formulated and optimal solvents to achieve highest reaction rates are identified. For considering the multiple and complicated effects of solvents on a chemical process, an integrated solvent and process design is performed. Solvent molecular structures and process operations are simultaneously optimized by the formulation and solution of a mixed‐integer nonlinear program. The proposed design methodology is exemplified for a selected Diels–Alder reaction. © 2014 American Institute of Chemical Engineers AIChE J, 61: 147–158, 2015