Molecular　docking　is　one　of　the　most　important　methods　in　computer　assisted　drug　design,　with　the　dreg　activity　of　flexible　ligands　verified　during　docking　to　macromolecular　targets,　its　use　value　could　be　determined.　Because　of　the　large　quantity　of　ligands　in　data　bank,　it　will　be　extraordinarily　time-consuming　to　compute　the　interaction　between　ligands　and　macromolecules.　The　paper　lays　emphasis　on　Lamarckian　Genetic　Algorithm　of　Autodock,　furthermore,　in　accordance　with　the　key　links　in　the　Autodock　program,　a　suite　of　parallel　approaches　is　not　only　designed　but　implemented.　In　the　end,　the　parallel　approaches　prove　effective　in　improving　the　performance　of　Autodock　through　thoroughly　testing.　©　2008　IEEE.