A pilot scale two-stage batch diafiltration process for whey protein concentrate (WPC) production is presented in this work. This process has two main advantages: a significant water saving with respect to a single-stage diafiltration process and a membrane surface saving with respect to a continuous multistage process. Every unit operation of the process has been experimented in a pilot scale (ultrafiltration, diafiltration, drying), in order to produce a WPC powder. Lactose content decreased from about 75% (of whey) to 4.5% (calculated as mass of lactose per total solute mass) and proteins increased from 15% to 83% (calculated as mass of proteins per total solute mass), with a water consumption of about 1.5 L/L of whey. Permeability tests enabled the calculation of the main mass-transfer resistances through the membrane: the intrinsic membrane resistance, R-M, was 13.8 +/-0.1, the one due to fouling, R-F, was 3.68 +/-0.03, and the one due to polarization, R-G, was estimated at 2.4 +/-0.4 (at 150 kPa TMP) 10(9) Pa s m(-1), at 25 degreesC. These values are in agreement with data reported in the literature, A study performed on the membrane retention properties showed that not only peptides but also two native proteins (beta -lactoglobulin and alpha -lactoalbumin) pass in the permeate during ultrafiltration. Consequently, a significant protein loss takes place during the process. Diafiltration tests performed in a two-stage countercurrent operation demonstrated the water saving with respect to a single-stage process (46%). Furthermore, experimental data have been used in order to evidence the capability of a previously developed mathematical model to predict the components' concentration during diafiltration.