An　intelligent　algorithm　based　on　genetic　algorithm　search　for　selecting　an　optimal　gene　subset　was　presented　and　applied　it　to　the　Chinese　gastric　cancer　mRNA　microarray　data　to　discover　the　feature　genes.　Firstly,　an　improved　genetic　algorithm　was　proposed　to　learn　the　optimal　subsets　of　genes.　Then　a　support　vector　machine　(SVM)　was　employed　to　find　the　gene　subset　with　best　classification　performance　for　distinguishing　cancerous　tissues　and　their　counterparts.　Some　of　the　obtained　feature　genes　have　been　validated　by　Beijing　Molecular　Oncology　Laboratory.　Both　of　the　biological　and　computational　experiments　have　shown　that　the　improved　genetic　algorithm　has　good　performance　in　both　the　quality　of　obtained　feature　subsets　and　computation　efficiency.　©　2007　IEEE.