We present a novel approach to the protein sequence design problem. Given a target native structure and a target temperature (T-D), our method generates sequences for which the target structure is most likely to be the lowest energy structure. Furthermore, the target structure is rapidly reached at T-D. Thus, the simultaneous requirements of stability and kinetic accessibility are satisfied in our design algorithm. The method consists of optimizing a function that captures the energy-landscape features responsible for the native state stability and folding kinetics of protein-like heteropolymers. The efficacy of our method is demonstrated by applications to lattice models (with and without side chains) in which the interaction energies involve pair potentials. The optimization process is computationally efficient. Optimal sequences satisfy the relation T-D approximate to T-theta, where T-theta is the collapse transition temperature.