Logbook data for the Taiwanese Pacific saury fishery and multi-sensor satellite images for 2006–2010 were used to characterize the habitat of Pacific saury (Cololabis saira) in the northwestern Pacific (NWP). An empirical cumulative distribution function (ECDF) approach identified that high cpue (catch per unit of effort) of Pacific saury occurred when sea surface temperature (SST) ranged from 14 to 16°C, chlorophyll-a concentration (Chl a) ranged from 0.4 to 0.6 mg m−3, and net primary production (NPP) ranged from 600 to 800 mg C m−2 d−1. A generalized additive model (GAM) and spatial non-stationary geographically weighted regression (GWR) were applied to predict the habitats of Pacific saury in the NWP. The spatial distributions of Pacific saury in the NWP estimated using the two approaches were similar and matched the nominal cpue distributions and those inferred from preferred habitat ranges based on ECDF. The density of Pacific saury is higher in coastal waters close to the island of Hokkaido and near the southern Kuril Islands than in the open sea. SST, Chl a, and NPP were substantially higher in the fishing grounds for Pacific saury during the main fishing season (September and October), corresponding to a high cpue for Pacific saury (23.1 t fishing day−1). The GAM explained more variability in spatial distribution (35.7%) than GWR (20.5%) VGPM (Vertically Generalized Production Model). Results derived from this study could improve our understanding of Pacific saury habitat distributions, which could be used to forecast fishing grounds and to develop fishery management advice based on oceanographic conditions that might be impacted by climate change.