The model of the pore geometry of sandstones proposed in this paper is based on three-dimensional qualitative observations of the pore structure of sandstones reported in the literature. Three categories of pores can be distinguished: large spherical pores at four-grain vertices (or nodal pores); tube-like throats at three-grain edges; and narrow sheet-like throats at two-grain faces. As a first step, the model is limited to rocks with a low clay content and, therefore, pores and throats are supposed to be free of clay. In this model, the sheet-like and tube-like throats are in parallel and, because of their different shapes, respond differently to pressure. The transport properties can thus be split into two parts: a strongly pressure dependent part associated with the sheet-like throats and an approximately constant part associated with the tube-like throats and the equidimensional nodal pores. Following these remarks, a scheme to estimate the volume fraction of each category of pores using the pressure dependence of the transport properties and a simple capillary model is developed. This approach was applied to data from the literature and appeared to succeed and fail when it was expected to (however, one exception was noted). For 15 of the 27 sandstones considered, the volume fraction and the size of the throats and pores predicted agreed in order of magnitude with microstructure data gathered in the literature. For several of the 12 remaining sandstones, independent microstructure data showed that basic assumptions of the model were not satisfied. Eight of these sandstones were associated with high clay content, and it was speculated that the last four might contain semimacroscopic fissures.