Bull. Jpn. Soc. Fish. Oceanogr. 68(2), Page 106-120, 2004   A review of some statistical approaches used for CPUE standardization Hiroshi Shono National Research Institute of Far Seas Fisheries, Fisheries Research Agency, 5-7-1 Shimizu-Orido, Shizuoka-shi, Shizuoka 424-8633, Japan e-mail: hshono@affrc.go.jp This paper outlines some approaches to CPUE standardization using statistical modeling, data mining technique and describes several proper problems of CPUE analyses in the fish population dynamics. In the statistical modeling, we mainly concentrate on the generalized linear model which is widely utilized for CPUE standardization. We describe the two typical fixed models (CPUE model and catch model), mixed model using random effect. The problems of variable selection and model comparison are also discussed. In the approach for data mining, this paper describes the applications of tree regression models, neural networks and generalized additive models to CPUE standardization as an alternative to statistical way. We mainly focus on the procedure to factorial experiment including the extraction of CPUE year trend and predict CPUE corresponding unsupervised data by these methods for data mining. As the proper problems of CPUE analyses, we express the problems of so called zero-catch, the definition of abundance index and other some issues such as habitat model and the handling of fishing effort in the purse seine fisheries. Key words: CPUE standardization, generalized linear model, random effect, model selection, data mining, tree regression model, neural networks, generalized additive model, zero-catch problem, abundance index, habitat model