alpha-fetoprotein　(AFP)　is　a　useful　diagnostic　marker　for　the　detection　of　a　great　number　of　infantile　diseases,　especially　for　some　forms　of　malignant　tumors　and　liver　disorders.　Although　there　are　several　methods　for　quantitative　determination　of　AFP,　compared　with　these　methods,　the　gold　immunochromatographic　assay　(GICA)has　the　advantages　of　easy　to　handle,　low　costs,　quick,　no　requirement　for　skilled　technicians　and　most　important,　well　suitable　for　point-of-care　measurements.　However,　GICA　mostly　has　been　used　for　qualitative　or　semi-quantitative　detection　visually.　This　paper　presents　a　rapid　quantitative　determination　method　of　AFP　with　GICA　strip　based　on　the　reflective　optical　detection.　Under　the　driving　of　the　micro-stepper　motor,　we　can　get　a　curve　of　GICA　strip　signal　distribution.　Then　after　segmentation　of　the　test　line　and　control　line　using　fuzzy　c-means　clustering　algorithm,　the　features　were　extracted　to　be　as　the　input　features　of　Support　Vector　Regression　(SVR)　model.　SVR　was　used　for　prediction　of　the　AFP　concentration.　To　evaluate　the　reliability　and　accuracy　of　GICA　for　fast　quantitatively　determination　of　AFP　and　its　clinical　value,　the　AFP　concentration　of　specimen　was　tested　by　GICA　and　chemiluminescent　immunoassay　analyzer,　and　analyzed　the　coincident　rate　and　correlation　of　these　two　methods.　The　experiment　results　demonstrated　that　the　correlation　coefficient　results　between　the　two　methods　was　0.961.　The　test　results　showed　that　the　SVR　model　yields　a　good　result　and　is　proved　to　be　appropriate　in　quantitative　determination　of　AFP　with　GICA　strip.　©　2012　Academy　Publisher.