A comparison is made between the conventional macroscopic pore theory, the single-file (no-pass) theory, and the bimodal theory in their ability to predict the values of the unit osmotic permeability Pos of single pores. In larger pores osmosis is thought to be a viscous (bulk) flow, while in molecular-sized pores only diffusive flow is considered possible. The physical assumptions underlying these theories are examined and compared with bimodal theory in which (i) viscous flow is impossible in any pore region that can be permeated by the driving osmolyte, and (ii) a distinction between diffusive and viscous flow can still be present in no-pass pores. Experimental values for the osmotic permeability of channels formed by the antibiotics amphotericin, nystatin, and gramicidin and the cellular aquaporin CHIP28 determined with different osmolytes are compared with theoretical expressions for Pos as a function of osmolyte radius. Aquaporins are probably pores of variable internal cross section and bimodal theory predicts that they can be probed by osmolytes of different radius to give different osmotic flows, although the overall permeability to each molecule is apparently zero. Such information can be used to construct a model of the pore channel. Conversely, if the pore structure is known, the unit osmotic permeability to any osmolyte can be calculated.