The　GPS　/　INS　integrated　navigation　system　performance　will　significantly　decrease　during　GPS　outages.　In　this　paper,　we　study　an　new　integrated　navigation　algorithm　based　on　Adaptive　Neural-　fuzzy　Inference　System　(ANFIS).　The　algorithm　adopted　Kalman　filter　with　pseudo-range　and　pseudo-range　rate　observations　when　the　number　of　GPS　satellites　was　not　less　4.　Otherwise,　ANFIS　was　used　to　estimate　the　navigation　errors　and　restrain　the　increasing　INS　errors　to　achieve　integrated　navigation.　The　new　algorithm　can　improve　the　performance　of　integrated　system　effectively　and　enhance　the　horizontal　position　accuracy　than　traditional　tight　integration　algorithms.　Especially,　the　method　is　applicable　to　the　complex　work　environment　of　navigation　systems　of　ships.