Estimating chemical equilibrium constants, reaction rates, and branching ratios using computational chemistry requires methods for locating stationary points on potential energy surfaces. I will review our recent work on developing computational methods for determining chemical reaction pathways and refining the structures of molecular reactants, products, transition-states, and stable intermediates in chemical reactions. Much of our work exploits the realization that, in elementary chemical reactions, only a few chemical bonds are broken and formed. By focusing our attention on these key coordinates and approximating the remainder of the molecule as a bath, we can efficiently and reliably characterize key points on chemical reaction pathways. These methods are useful for reactions in any context, and we have tested them for reactions relevant to environmental chemistry, materials science, and chemical biology. If time permits, I will also present a few other key optimization problems in computational quantum chemistry, including special cases of large-scale semidefinite programming and nonlinear least-squares.