1. Oral loads have been used to assess the permeability of the human gastrointestinal tract, with lactulose (mol. wt. 342), raffinose (mol. wt. 504), stachyose (mol. wt. 666) and a fluoresceinlabelled dextran (mol. wt. 3000) as marker substances. Timed urinary recovery of these substances, which are not metabolized, was measured by quantitative paper chromatography and direct fluorimetry, and the results were used as an indication of passive intestinal permeability. 2. Results in healthy adults showed that permeability to these markers was dependent on molecular size, even after correction for aqueous diffusion differences, such that a profile of restricted permeability could be described for this range of markers. Interpretation in terms of conventional pore theory suggested the presence of more than one population of pores. 3. Ingestion of solutions made hyperosmotic by inclusion of glycerol resulted in a large increase in permeability, in a pattern that suggested an increase in either the size or frequency of a range of smaller pores. 4. A similar increase in permeability and alteration in the profile of restriction was found in patients with coeliac disease. 5. The possible location of such pores in the gastrointestinal mucosa is discussed in relation to the cell membrane, the intercellular junction, and the sites of cell exfoliation.