Bull. Jpn. Soc. Fish. Oceanogr.  68(1), Page 44-51 , 2004   Statistical Test and Model Selection of Fish Growth Formula Tatsuro AKAMINE National Research Institute of Fisheries Science, Fisheries Research Agency, 2-12-4 Fukuura, Kanazawa-ku, Yokohama, Kanagawa 236-8648, Japan e-mail: akabe@affrc.go.jp Many formulae have been presented for fish growth. Althougth Schnute (1981) formula is widely used in fish population dynamics, it is equivalent to Richards (1959) one. Akamine-Richards formula presented by Akamine (1993) is the simplest and the most practical one which integrates von Bertalanffy, its cubic, Gompertz, logistic, and their seasonal oscillating models expanded by using a periodic function. In this paper, we discuss statistical tests and interval estimations of parameters for fish growth formulae. In particular, the weighted least-squares method for curve fitting, chi-square test for the data with variance, F-test for the data without variance, and model selection using AIC (Akaike Information Criterion) are explained mathematically, and a numerical example of interval estimations for von Bertalanffy formula is presented by using a spreadsheet program. Key words: growth, interval estimation, model selection, statistical test