Abstract: An optimal conformation search approach to the rigid receptor and flexible ligand docking problem is presented. The conformation searching process in molecular docking is modeled as an optimization problem with multiple constraints, and a new iteration scheme in conjunction with multi-population genetic strategy and an entropy-based searching technique is developed to do the conformation optimization. In the proposed GA, the general problem of early premature is avoided by the diversity of populations; taking the precision of contracted variables space as terminate rule makes the algorithm convergent reasonably; the specific strategy of reserving the most fitness member with evolutionary historic information ensures the global solution; the application of micro-population cuts down the calculation complexity. Besides, a parallel computing strategy with coarse grain is also employed on a Tianchao Dawning 3000 parallel computer and a high speedup and efficiency are reached. An application of the proposed method is illustrated.