Objective: Whether or not there is significant paracellular water permeability in leaky epithelia is still unresolved because it is difficult to experimentally separate transcellular from tight junction (TJ)- controlled paracellular water flux. As a novel approach, we measured transepithelial water flux with and without selective molecular perturbation of the TJ to univocally attribute changes to the paracellular pathway. Methods: MDCK C7 kidney tubule cells were stably transfected with either claudin-2 or claudin- 10b, two paracellular cation channel-forming TJ proteins, which were not endogenously expressed. Development of a specific modified Ussing chamber allowed quantification of water flux. Water flux was induced by an osmotic gradient, a Na ⁺ gradient, or both. Results: Transfections did not alter the expression of endogenous claudins or aquaporins. Application of an osmotic gradient, a Na ⁺ gradient, or both resulted in an increased water flux in claudin-2-transfected cells compared to vector controls, indicating claudin-2-mediated paracellular water permeability. Na ⁺-driven water transport in the absence of an osmotic gradient indicated a single-file mechanism. In contrast, claudin-10b transfection did not alter water flux. Conclusion: We demonstrate that claudin-2, but not claudin-10b, forms a paracellular water channel and by this mediates paracellular water transport in leaky epithelia. This is in accordance with the physiological role of these TJ proteins, as claudin-2 is typical for leaky, water-transporting epithelia, and claudin-10b is expressed in water-impermeable segments of Henle's loop.