In terms of performance measurement on Recommendation Systems, previous research focuses on system viewpoints. For Example, Leng¡¦s NewsWeeder(1995) measure recommendation performance by precision¡FSyskill&Webert measure recommendation performance by classification accuracy¡FGroupLen measure recommendation performance by system response time¡CWe bring up a user-oriented viewpoint which means that whether the recommendation interface satisfies user¡¦s needs, whether it is easy to use, and whether it provides sufficient information to user. In the meantime, prvious research didn¡¦t think over the difference of everyone¡¦s information processing style¡CTherefore, our research objective focuses on effect of cognitive styles and recommendation style on user satisfaction¡C In the construct of recommendation interface, we adopt average rating and text comment¡CAnd in the construct of cognitive style, we classify it with intuitive and analytical users¡CThe measurement of user satisfaction adopts Doll and Torkzadea (1988) questionnaire and refines it¡C The research result finds that different recommendation interfaces and cognitive styles have a significant impact on user satisfaction¡CIf we don¡¦t think over effect of cognitive styles, there is higher user satisfaction on text comment¡CIf we think over effect of cognitive styles, intuitive user has higher user satisfaction on average rating¡Fanalytical user has higher user satisfaction on text comment¡COur research contribution is as follow¡CIn academic aspect, our research finding can provide researcher in cognitive psychology¡Binformation recommendation and information management field for further research¡FIn practical aspect, our research finding can assist webstore company in implementing recommendation service¡C