Water-fat MRI tissue composition can be described by a three compartment model, i.e. water, fat and MR-invisible tissue. The water intensity (W) is proportional to the water proton density (PD) and fat (F) to the lipid PD. If a compartment’s fat content is known and invisible volume is constant, then W and lean tissue volume should be proportional to the hydration. It should thus be possible to determine the hydration level of a lean tissue based on W.11 hemodialysis patients were recruited, whole-body T2* and lipid spectra compensated water-fat MRI images were collected pre/post dialysis. The net fluid drawn (NFD), i.e. machine setting compensated for tubular dead space and ingestion, was logged. Images were acquired on a Philips Ingenia 3T with 10 axial 3D Spoiled GRE stack with alpha=10, TE=1.15, 2.3, 3.35, 4.6 ms, TR 5.8 ms, FOV 340x560 mm2, voxel size 2.5x2.5x4 mm3.Leg and abdominal muscles were segmented automatically (Karlsson, jMRI 2015). W and F were calibrated using adipose tissue (AT) as an intensity reference (Romu, ISBI 2011), so F becomes the AT concentration and W is related to the PD of lipids in AT. Based on the average W of muscle tissue a hydration ratio H = W/(t*g) was computed; t=1+(f-fn) compensates for the hydration of AT, f is the AT fat fractions (F/(F+W)) of the patient and fn the normal value; g=k*F+m compensates for fat infiltration, F is the average muscle AT concentration, m and k are set such that g=H in normally hydrated tissue. Thus, H=1 in normally hydrated tissue. Determining fn, m and k is beyond the scope of this study, fn=0.95, k=-0.6 and m=0.65 was used. The lean muscle volume (V) and muscle H differences was measured pre/post, and tested by paired T-tests.