Discovering　service-on-demand　for　large　numbers　of　functionality-similar　web　services　is　one　of　the　key　issues　in　service　discovery　study.　To　find　out　the　proper　service　among　the　functionality-similar　web　services,　a　merging　cluster　algorithm　regarding　QoS-oriented　supply　and　demand　is　proposed　in　this　paper.　To　meet　the　target,　FCM　clustering　is　adopted　for　agglomerative　clustering　between　the　user＇s　QoS　requirement　information　and　the　QoS　information　from　Web　Service　resources.　Then,　the　sequence　could　be　determined　by　similarity　computation　with　the　same　clustering.　Lastly,　the　numerical　example　is　presented　to　illustrate　that　the　service-on-demand　can　be　discovered　efficiently　and　effectively.