Abstract: A class of bilevel multl-objective programming was converted into the problem of equivalent single-level multi-objective programming. Then a new chaos genetic optimization algorithm was presented by using the inversion property of genetic algorithm and the ergodic property of chaos optimization method and combining with the exact l₁ penalty function. The local search ability and search accuracy of genetic algorithm were improved. The solving accuracy and credibility became high. An actual calculated example showed that the algorithm is effective and efficient.