This paper proposes an algorithm that estimates blood viscosity during cardiopulmonary bypass (CPB) and validates its application in clinical cases. The proposed algorithm involves adjustable parameters based on the oxygenator and fluid types and estimates blood viscosity based on pressure-flow characteristics of the fluid perfusing through the oxygenator. This novel nonlinear model requires four parameters that were derived by in vitro experiments. The results estimated by the proposed method were then compared with a conventional linear model to demonstrate the former's optimal curve fitting. The viscosity (ηe) estimated using the proposed algorithm and the viscosity (η) measured using a viscometer were compared for 20 patients who underwent mildly hypothermic CPB. The developed system was applied to ten patients, and ηe was recorded for comparisons with hematocrit and blood temperature. The residual sum of squares between the two curve fittings confirmed the significant difference, with p < 0.001. ηe and η showed a very strong correlation with R2 = 0.9537 and p < 0.001. Regarding the mean coefficient of determination for all cases, the hematocrit and temperature showed weak correlations at 0.33 ± 0.14 and 0.22 ± 0.21, respectively. For CPB measurements of all cases, ηe was more than 98% distributed in the range from 1 to 3 mPa⋅s. This new system for estimating viscosity may be useful for detecting various viscosity-related effects that may occur during CPB.