Nanofluids viscosity is one of the most important thermophysical properties in nanofluids usage especially in chemical and petroleum engineering applications. So it is highly desirable to predict the viscosity of nanofluids accurately. Experimental measurements are impossible in most situations and present models are not comprehensive and efficient especially for high temperature, high volume concentration and high viscosity values. In this study, a new correlation has been developed based on the comprehensive database of water based Al₂O₃, TiO₂, SiO₂ and CuO nanofluids viscosity data found in literature. The proposed correlation uses temperature, nanoparticle size, nanoparticle volumetric concentration and water viscosity as inputs and gives the viscosity of nanofluid as output. The results demonstrated that predictions obtained by this new correlation are in a better agreement compared with experimental values rather than the previous models especially for high temperature, high volumetric concentration and high viscosity values. This superiority is also preserved in trend predictions with temperature and volumetric concentration.