Outflow obstruction of the submandibular duct, for a short period during parasympathetic secretion, caused an increase in glandular permeability to horseradish peroxidase. Higher frequencies of parasympathetic nerve stimulation during the obstructive period induced greater increases in the permeability. It is likely that the intraluminal distensions had disrupted some tight junctions and so permitted a greater paracellular leakiness. The damage tended to induce ballooning between striated ductal cells, similar to the appearances observed by Emmelin, Garrett & Gjörstrup (1977 a) when sympathetic secretion was accompanied by myoepithelial contraction against a raised outflow pressure. The present results have been compared with experimental data from other workers and give support to the idea that tight junctions can behave in a sieve-like manner towards the back-diffusion of molecules across them. This work indicates that there is a great need for care in permeability studies. It is possible that some permeability changes observed by other workers may have been the consequences of physical damage between cells. Our findings reinforce the need for morphological assessment of the glands after permeability experiments. In man it is likely that naturally occurring or artificially induced obstructive events may at times create similar permeability changes and these may enable protective substances to pass from the blood to the saliva. This affords possible explanation for such phenomena as the therapeutic value that often accrues from sialography and the efficacy of many different antibiotics in obstructive sialadenitis, despite the fact that most antibiotics do not normally permeate to the saliva.