To　find　an　optimal　multi-machine　scheduling　for　objective　tasks　with　deadline　constraints,　an　optimal　model　was　proposed,　and　GASA　hybrid　optimal　strategy　was　applied　to　solve　this　problem.　Each　individual　has　two　gene　clusters,　one　record　the　order　of　the　tasks　to　be　executed,　the　other　stands　for　the　number　of　the　tasks　allocated　to　each　machine.　Individuals　created　by　greedy　algorithm　were　introduced　to　improve　adaptability　of　initial　population,　and　simulated　annealing　algorithm　was　introduced　to　avoid　prematurity.　Several　simulation　experiments　show　that　the　proposed　scheduling　algorithm　is　valid　and　feasible.　©　2006　IEEE.