Abstract: Computer simulation has been developed into an important tool for the elucidation of biological functions from the atomic level structures and dynamics of biomolecules. Compared with current experimental techniques probing atomic level structures, simulations provide not only averages, but also distributions. Besides structures, biological functions often rely on sophisticatedly controlled dynamics of biomolecules, such as the allosteric effects in enzyme catalysis or the effects of ligand-receptor binding in signal transduction. To help understand and eventually control such processes, simulations can be used to reconstruct conformational pathways, identifying intermediates and transition states. This report highlights our recent work in this field. One focus of our research is on developing, testing and refining energy functions for protein simulations, including pure molecular mechanical models for modeling conformational dynamics and hybrid quantum mechanical/molecular mechanical models for modeling enzyme catalysis. Another focus is on developing methods for efficient sampling in the conformational space and for mapping conformational pathways.