Abstract: Ommastrephes bartramii is one of the most important commercial fishing targets for China in the Northwest Pacific. Accurate forecast of the fishing ground helps locate the shoal, save fuel and improve yield. In the present study, historical catch log data and environmental factors including sea surface temperature (SST) , chlorophyll-a concentration (Chl) , SST gradient (SSTG) and Chl gradient (ChlG) in the Northwest Pacific Ocean were collected. Based on the support vector machine (SVM) , squid fishing ground forecast model was established with Radial Basis Function (RBF) kernel in the monthly resolution and the spatial resolution of 0.5 0.5. The optimal combination of penalty parameter (C=1.41) and kernel parameter (=2.83) were obtained by 10-fold cross validation and grid-search when the accuracy of model reached 73.6%. A simulated accountancy testwas carried out usingmonthly environmental data in 2013. The accuracy rate ranged from 53.4% to 60.0% with 57.4% on average. The result suggests that SVMcan provide an efficient means for squid fishing ground forecast with a small training dataset in the Northwest Pacific.