Electrolyte concentration values in interstitial fluid samples that have been reported by a number of authors were markedly different from those of a hypothetical ultrafiltrate of plasma. Because no adequate explanation has been provided for the discrepancy, we attempted to study the question 1) by measuring ion and protein concentration in the plasma and in the interstitial fluid samples, and 2) by constructing a theoretical model for ion distribution. Subcutaneous interstitial fluid samples were collected in rats by the implanted capsule and by the liquid paraffin cavity techniques. The samples were analyzed for sodium, potassium, calcium, chloride, total protein, and protein fractions. The ion distribution between vascular and interstitial compartments was found to correspond to the Donnan equilibrium. On theoretical ground it was concluded: the Donnan distribution is valid, if the size of "free-fluid spaces" is relatively large (r greater than 0.03 micron) compared with the rather short range of electrostatic interactions (approximately 0.8 nm). Due to the relatively small difference in protein concentrations between blood plasma and interstitial fluid and to the short range of electrostatic interactions, the influence of proteins on the distribution of small ions is negligible.