The effect of membrane shear rate during nanofiltration of skim milk diluted with two volumes of water representing dairy effluents, using a vibrating VSEP module with a Desal 5 DK membrane, was investigated. This shear rate (γ max ) was varied by performing tests at 60.75, 60.2, 60.0 and 59.8Hz. In long-term industrial use, it is necessary to limit the vibration amplitude d of the membrane rim to about 2–2.3cm, against d=3.0cm at the maximum and resonant frequency of 60.75Hz. At a pressure (TMP) of 4000kPa, 45°C and initial concentration, permeate fluxes decreased from 220Lh −1 m −2 at 60.75Hz (with a shear rate of 1.35×10 5 s −1 ) to 95Lh −1 m −2 at 59.8Hz (0.64×10 5 s −1 , d=1.5cm). The permeate carbon oxygen demand (COD) due to lactose increased from 20mgO 2 L −1 at 60.75Hz to 60mgO 2 L −1 at 59.8Hz, remaining much smaller than in the feed (36,000mgO 2 L −1 ). Permeate conductivity increased from 500μScm −1 at 60.75Hz to 800μScm −1 at 59.8Hz, for a feed conductivity of 2000μScm −1 . During concentration tests, performed at 4 frequencies, the permeate flux J at various volume reduction ratios of 1, 3 and 7 was given by a single equation J=3.0×10−6 γmax1.56, showing that the flux is mainly controlled by shear rate whether this shear rate is lowered by reducing the frequency or increasing the concentration. Tests performed with a real dairy effluent gave a variation of permeate flux and conductivity with TMP similar to the model, up to 3000kPa. Above 3000kPa, the real effluent flux dropped due to higher membrane fouling.