The Hydraulic Institute method, developed for predicting centrifugal pump performance of viscous Newtonian fluids, is used by some for non-Newtonian fluids. This requires an average value of viscosity representing the variable non-Newtonian viscosities. To determine such an average viscosity, two approaches exist in the literature: the use of a Bingham plastic viscosity and the use of the apparent viscosity. Results from these two approaches are not in agreement. This study evaluates the two approaches using two independent datasets, obtained from three pumps and three fluids. Results indicate that using the apparent viscosity approach gave better head prediction and using the Bingham plastic viscosity resulted in better efficiency prediction.