A　model　of　artificial　neural　networks　(ANN)　and　genetic　algorithms　(GA)　are　used　for　the　analysis　and　prediction　of　the　correlation　between　brake　material　components　and　friction　performance.　After　the　ANN　model　based　on　BP　algorithms　is　trained　successfully,　genetic　algorithms　are　used　to　optimize　the　input　parameters　of　the　model　and　select　optimum　compositions　of　brake　material.　Here　the　input　parameters　of　the　artificial　neural　network　are　the　content　of　steel　fiber,　aluminosilicate　fiber,　phenolic　resin　and　Cu　powder.　The　outputs　of　the　ANN　model　are　one　of　the　friction　behaviors,　namely,　the　friction　coefficient　of　a　brake　material　and　the　wearing　rate.　The　optimum　compositions　(wt%)　of　the　brake　material　designed　by　ANN　and　GA　are　steel　fiber　32.2,　alumino-silicate　fiber　22.1,　phenolic　resin　28.9　and　Cu　powder　5.4.　Friction　coefficients　of　the　designed　brake　material　by　simulation　and　experiment　are　compared.　The　maximum　relative　error　is　5.3%,　whereas　the　minimum　is　0.3%.　The　results　of　the　comparison　of　the　actual　values　and　network　values　are　impressive.　This　indicates　that　the　neural　network　with　an　accuracy　that　is　acceptable　in　most　design　considerations　may　have　considerable　potential　for　solving　time-consuming　problems　for　the　design　of　the　automobile　brake　material.