Current-voltage relationships of a cation channel in the tonoplast of Beta vulgaris, as recorded in solutions with different activities of Ca2+ and K+ (from Johannes & Sanders 1995, J. Membrane Biol. 146:211-224), have been reevaluated for Ca2+/K+ selectivity. Since conversion of reversal voltages to permeability ratios by constant field equations is expected to fail because different ions do not move independently through a channel, the data have been analyzed with kinetic channel models instead. Since recent structural information on K+ channels show one short and predominant constriction, selectivity models with only one binding site are assumed here to reflect this region kinetically. The rigid-pore model with a main binding site between two energy barriers (nine free parameters) had intrinsic problems to describe the observed current-saturation at large (negative) voltages. The alternative, dynamic-pore model uses a selectivity filter in which the binding site alternates its orientation (empty, or occupied by either Ca2+ or K+) between the cytoplasmic side and the luminal side within a fraction of the electrical distance and in a rate-limiting fashion. Fits with this model describe the data well. The fits yield about a 10% electrical distance of the selectivity filter, located about 5% more cytoplasmic than the electrical center. For K+ translocation, reorientation of the unoccupied binding site (with a preference of about 6:5 to face the lumenal side) is rate limiting. For Ca2+, the results show high affinity to the binding site and low translocation rates (<1% of the K+ translocation rate). With the fitted model Ca2+ entry through the open channel has been calculated for physiological conditions. The model predicts a unitary open channel current of about 100 fA which is insensitive to cytoplasmic Ca2+ concentrations (between 0.1 and 1 microM) and which shows little sensitivity to the voltage across the tonoplast.