The　classification　of　the　grade　of　shrink　and　expansion　for　the　expansive　soils　is　very　important　work　for　constructions　in　expansive　soil　area.　Accurate　evaluation　of　the　estate　of　expansive　soil　may　provide　scientific　data　and　basis　for　the　prevention　of　engineering　design　and　construction.　This　paper　brings　forward　the　classification　method　for　expansive　soil　based　on　fractal　interpolation　theory.　Firstly,　five　indexes,　including　the　liquid　limit,　the　overall　rate　of　expansion　and　contraction,　plastic　index,　moisture　content,　free　expansion　rate,　are　selected　as　the　factors　for　synthetic　evaluation　of　expansive　soil.　In　addition,　because　of　few　standard　samples,　forty　standard　samples　are　produced　by　the　way　of　stochastic　interpolation　on　basis　of　classification　standard.　Secondly,　through　calculating　the　fractal　dimension　of　single　index　of　expansive　soil,　confirming　the　fractal　dimension　exponent　evaluation　of　the　index　of　single　expansive　soil　with　the　maximum　likelihood　classification　principle.　Thirdly,　adopting　weighted　summation　value　this　method　calculates　the　level　of　the　colligate　estimation　of　multi-index　of　expansive　soil　and　establishes　the　fractal　interpolation　evaluation　model　according　to　the　relationship　between　the　value　of　comprehensive　evaluation　of　the　samples　and　the　experience　level.　Finally,　several　samples　are　classified　based　on　the　model.　The　classification　results　are　consistent　with　other　methods　and　show　that　the　model　we　made　is　simple,　easy　to　solve　the　problem,　and　can　get　accurate　solutions,　and　then　could　be　used　widely　in　classification　of　expansive　soil.　Moreover,　particular　score　of　each　sample　could　be　computed　by　the　model;　so　it　has　more　precise　classification　and　is　a　rather　better　new　method.　It　is　possible　to　make　use　of　fractal　interpolation　model　to　evaluate　the　levels　of　expansive　soil.　And　it　can　provide　reference　for　finding　parameters　in　engineering　design　and　calculation.　©　2014　ejge.