Studies of the transcapillary exchange of fluid and solutes have provided experimental evidence for the following description of the capillary wall: Transport can be adequately described by passive phenomena such as filtration and diffusion across the permeable structures of the capillary barrier. Hydrophilic solutes are progressively restricted in their transcapillary passage with increasing molecular radius in a bimodal manner. The simplest membrane model compatible with these properties is the two-pore model (Grotte 1956), for which there now is massive documentation (Taylor & Granger 1984). Thus, small hydrophilic solutes (radius less than 30A) are transported mainly by diffusion through small equivalent pores (40-65A), which also represent almost 90% of the hydraulic conductivity. Moreover, larger solutes pass mainly through large equivalent pores (250-350A) by convection. Hence, diffusion is of minor importance for macromolecular transport also during conditions of no net fluid flux across the capillary walls, when there is a circulation of fluid between small and large pores and a net filtration of macromolecules at the large pores. The functional small pores are most probably identical to the interendothelial junctions, while the large pore system is more difficult to define. In addition, solutes are subjected to electrostatic charge interactions at the capillary wall. Thus, there is overwhelming evidence of the presence of high densities of negative charges at the capillary endothelium (glycocalyx), the basement membrane and in the interstitium. This suggests the presence of a negative capillary charge barrier restricting anionic solutes, which also could be experimentally verified (study I & II), but the importance of this barrier is not yet fully clear. A simplistic theoretical two-pore model including effects of charge (Munch et al. 1979) was found to describe our experimental data. According to this model, the charge effect will be most important for solutes with molecular radii of 20-40 A, while charges probably are of no importance for the transport of larger solutes. For anionic macromolecules, e.g. albumin, the effective pore radius will be 40-45A, but the steric small pore radius (i.e. for neutral solutes) will be around 65A. Hereby, the area of diffusion calculated from CFC and the small pore radius (64A) will be lower and it will actually approach the values determined by indicator dilution technique (cf. Haraldsson & Rippe 1986).(ABSTRACT TRUNCATED AT 400 WORDS)