The DSPM (Dorman steric partitioning pore model) was evaluated in the case of a titania membrane with "nanofiltration properties" by measuring the electrokinetic charge, pore size, and water permeability of the membrane, along with charged and uncharged solute retention. The zeta potential values (zeta) were determined from measurements of the electrophoretic mobility (EM) of titania powder forming the filtering layer of the membrane. Zeta potential values were converted into membrane volume charge (X) by assuming two limiting cases: a constant surface charge (sigma(cst)(s)) and a constant surface potential (psi(cst)(s)). The mean pore radius and thickness/porosity ratio of the membrane were determined by permporometry and from water permeability measureme ms, respectively. Retention measurements were carried out as a function of the permeate volume flux for both neutral solutes (polyethylene glycol PEG of different size) and salts (KCl, MgSO4, K2SO4, and MgCl2) at various pH values. Ionic retentions showed minimum values near the IEP of the membrane. Retention data were analyzed using the DSPM. Very good agreement was found between the pore radius calculated by the model and that determined by permporometry. X values calculated from fitting retention data using the DSPM were also in satisfactorily agreement with X values calculated from EM measurements assuming a constant surface potential for a large pH range. Furthermore, the DSPM leads to X values (X-DSPM) between those calculated from EM (X-EM) using the two limiting bounds. In other words, X-DSPM was higher than X-EM assuming psi(cst)(s) at pH values far from the isoelectric point (IEP) and lower than XEM assuming "cst, These results show that the DSPM is in qualitative agreement with the charge regulation theory (increase of the pore surface potential and decrease of the pore surface charge density with decreasing the pore size). On the other hand, the thickness/porosity ratio of the membrane calculated from solute retention data differed significantly from that determined from water permeability measurements. Moreover, a single value of Deltax/A(k) could not be determined from PEG and salt retention data. This means that the Deltax/A(k) parameter loses its physical meaning and includes physical phenomena which are not taken into account by the DSPM. Nevertheless, the model satisfactorily predicted the limiting retention, as this is not influenced by the Deltax/A(k) parameter. (C) 2003 Elsevier Science (USA). All rights reserved.