Based on shield driving data at Chongqing Yangtze River tunnel and fuzzy diagnosis principle，rock cuttabilities are studied. Point load strength tests of representative mudstone and sandstone in field are carried out. Thrust force and penetration rate(PR) of shield corresponding to the representative mudstone and sandstone are recorded by data acquisition and storage system of shield. Based on multivariate regression analysis，corresponding prediction models of penetration rate are developed，and then the attribution spaces of shield penetration rate are constructed and divided according to rock properties. Rock cuttabilities are flagged by corresponding PR of shield. Making use of fuzzy diagnosis principle，difference between PR in a rock sample and PR in representative rock is defined as a conception of distance. The correspondence between PR and normalized distance is developed. Membership of PR in a rock sample to PR in representative rock is acquired subsequently. Consequently，recognition function for rock cuttabilities is constructed. The rocks along the tunnel are characterized as alternation of mudstone，sandstone and mudstone containing sandstone. These rocks are all subordinate to the range of soft rock to medium hard rock，and the scope of their hardness is not too wide. So the rock cuttabilities are divided into three levels as soft，medium soft and medium hard，based on which the rock cuttabilities along the tunnel is recognized. Results of the rock cuttabilities are beneficial supplements for engineering geology surveying reports during construction；and these results have provided scientific references for predicting the cutter wear and deciding the time of the cutter exchange. When the shield is advancing，accuracy control of shield has too been improved.