This study applied the generalized additive models (GAMs) to examine the relationship between sailfish catch per unit effort (CPUE) from 1998 to 2004 and environmental variables, which included satellite-derived remote sensing data and other spatio-temporal variables in the marginal seas of the northwestern Pacific Ocean. The results of the stepwise GAMs analysis found less important influences upon sailfish CPUE from chlorophyll-a concentration (Chl-a) and sea surface height anomaly (SSHA). The final GAMs fitting model was constructed by the spatio-temporal variables, including the year, month, latitude, longitude, sea surface temperature (SST), Pacific decadal oscillation (PDO), southern oscillation index (SOI), and the bathymetry of operational factors. There was a significant relationship (R2 = 0.52, P < 0.0000) among the predicted values of GAMs analysis and nominal sailfish CPUE. It showed that the relatively high sailfish CPUE was found in June and July and a higher sailfish CPUE between 28 °C and 30 °C SST. The highest value of sailfish CPUE was found in 29 °C SST. These results can benefit to examine progressively the feeding and spawning habitats and its possible migratory route of sailfish in the northwestern Pacific Ocean.