Treating the blood-tissue barrier as a two-pore membrane the separate fluid and solute fluxes occurring across 'small pores' and 'large pores' were modelled in continuous capillaries employing two-pore equations for the calculations together with the non-linear flux equation and theories for restricted diffusion and for the reflection coefficient (sigma). The two-pore equations derived proved useful for analyses of transvascular protein flux data obtained at low as well as at high filtration rates. These equations were applied to lymphatic protein flux data from dog paw (Renkin et al. 1977a, b) and to tracer albumin uptake data from rat skeletal muscle (Rippe et al. 1979). For both sets of data the small- and large-pore radii became closely similar, 44 vs. 45 A and 240 vs. 225 A, which also holds for the large-pore fractions of hydraulic conductivity (0.097 vs. 0.056). The main result of this analysis is that the passage of macromolecules normally occurring across the microvascular walls is almost entirely convective, and hence, dependent on the transmural hydrostatic and oncotic pressure gradients and on the hydraulic conductivity. For example, 75-90% of the transvascular passage of albumin was found to be due to convection through large pores at normal lymph flows, the remaining portion being mainly due to diffusion across small pores. Solutes larger than albumin were almost exclusively transported by convection across large pores. Two-pore heterogeneity was found to explain the previously observed variations of the apparent overall large solute diffusion capacity (PSapp) and the overall reflection coefficient (sigma f) with filtration rate and also previous overestimations of PS. Furthermore, the present results were not compatible with protein transport across any 'non-hydraulically conductive capillary pathways' as previously postulated from the lymphatic protein flux data analysed here (Renkin 1985).