Ions cross leaky epithelia such as the kidney proximal tubule both through the cells (transcellular route) and through the tight junctions (TJs) between cells (paracellular route) – but how does water cross leaky epithelia? Aquaporin water channels mediate the transcellular passage of water, but is there a significant paracellular water flux across these epithelia? On page 1913, Michael Fromm and colleagues report that the answer to this long-standing question is yes. The authors stably transfect MDCK C7 cells (a renal epithelial cell line with water-impermeable TJs) with either claudin-2 (a cation-channel-forming TJ protein expressed in leaky epithelia) or claudin-10b (also cation-channel forming, but present in the water-impermeable segments of Henle's loop) and measure the transepithelial water flux, through layers of the transfected cells, induced by an osmotic and/or Na+ gradient. Both claudins increase the cation permeability of the cells. However, the water flux in claudin-2-transfected cells is higher than in vector controls, whereas claudin-10b transfection does not alter water flux. Therefore, conclude the authors, the water permeability of TJs depends on their molecular composition, as claudin-2 but not claudin-10b forms a paracellular water channel.