The permeability of junctional complexes to ultrastructural tracers of different molecular weight and the freeze-fracture appearance of junctional structure were investigated in the resting and stimulated rat parotid gland. Tracers were administered retrogradely via the main excretory duct, and allowed to flow by gravity (16 mmHg) into the gland for 15-60 min. Secretion was induced in some animals by intraperitoneal injection of isoproterenol. In resting glands, the tracers microperoxidase , cytochrome c, myoglobin, tyrosinase (subunits), and hemoglobin were restricted to the luminal space of the acini and ducts. In glands stimulated 1-4 h before tracer administration, reaction product for microperoxidase , cytochrome c, myoglobin, and tyrosinase was found in the intercellular and interstitial spaces, whereas hemoglobin was usually retained in the lumina. In contrast, horseradish peroxidase and lactoperoxidase appeared to penetrate the tight junctions and reaction product was localized in the extracellular spaces in both resting and stimulated glands. Diffuse cytoplasmic staining for horseradish peroxidase and lactoperoxidase was frequently observed in acinar and duct cells. The distribution of horseradish peroxidase was similar in both Sprague-Dawley and Wistar-Furth rats, and at concentrations of 0.1-10 mg/ml in the tracer solution. Freeze-fracture replicas of stimulated acinar cells revealed an increased irregularity of the tight junction meshwork, but no obvious gaps or discontinuities were observed. These findings indicate that (a) tight junctions in the resting rat parotid gland are impermeable to tracers of molecular weight greater than or equal to 1,900; (b) stimulation with isoproterenol results in a transient increase in junctional permeability allowing passage of tracers of molecular weight less than or equal to 34,500; (c) junctional permeability cannot be directly correlated with junctional structure; and (d) the behavior of horseradish peroxidase and lactoperoxidase in the rat parotid gland is inconsistent with their molecular weights. Cell membrane damage due to the enzymatic activity or binding of these two tracers may account for the observed distribution.