Abstract: Brackish water, which typically represented the water quality in the Huang-Huai region along the junction of the north China piedmont alluvial plain and central alluvial plain, was simulated and used as feed water to run a nanofiltration (NF) membrane system. Bench-scale experiments were carried out to evaluate NF softening separating performance. The influences of transmembrane pressure (0.6-2.1 MPa), inlet tangential flow velocity (0.09-0.38 m/s) and feedwater temperature (7-35 ℃) on the softening efficiencies of NF membrane were investigated. The results revealed that NF system produced a highest flux of 52.04 L/(m ²·h) with Ca ²⁺, Mg ²⁺, CO_3^2- and SO_4^2- rejection of 44.13%, 73.72%, 81.05% and 99.13%, respectively, under the optimal operating conditions of transmembrane pressure at 1.2 MPa, inlet tangential flow velocity at 0.28 m/s and feedwater temperature at 15 ℃. Under the optimal operating conditions, NF membrane achieved the highest softening performance with 53.31% of the total hardness removal efficiency. Additionally, pH values of NF permeate decreased while pH of NF retentate increased gradually with increasing of transmembrane pressure, inlet tangential flow velocity and decreasing of feedwater temperature within the testing scope.