The aim of this study is to contribute to the sustainable utilization of yellowfin tuna (Thunnus albacares) resources. In this study, Generalized Additive Model (GAM) was applied to develop the integrated habitat index (IHI) models on the basis of the data which were collected in waters near Gilbert Islands from October, 2009 through December, 2009. The profiles of environmental variables and the yellowfin tuna catch rate data were collected at 34 sites. These models were used to predict the corresponding potential catch rate. The Wilcoxon test was used to test if there were significant differences between the predicted catch rates predicted by the models and the nominal catch rates. The integrated habitat index (IHI) in different water strata of yellowfin tuna were estimated by the predicted catch rate predicted by the models. Comparing the nominal CPUE and the average integrated habitat index (IHI) in different water strata, the predicted power of the models were evaluated. The correlation coefficients between the nominal CPUE and the average integrated habitat index (IHI) in different water strata were calculated. This correlation coefficient was also used to evaluate the prediction power of the models. In addition, the Gilbert Islands survey data in 16 sites from Nov. 2010 through Jan. 2011 were used to verify the effectiveness of the models. These data were input into the models for water stratum of 40-80 m and the whole water bin (0 ~ 240 m), and the IHIs of the yellowfin tuna were estimated. The results showed that: (1) The IHI distributions of different water stratum were different from each other. The environmental variables which influenced the distribution of yellowfin tuna were different from the different water strata. The yellowfin tuna mainly distributed in the water stratum of 40-120 m; (2) The prediction power of the models was good on the basis of the verification data obtained in 16 sites from Nov. 2010 through Jan. 2011; (3) GAM is suitable for the environmental variable selectivity, which influenced the distribution of yellowfin tuna, and shows the nonlinear relationship between the environmental variable and the catch rate; (4) GAM can be used to study the spatial distribution of the pelagic fish by building the IHI model.