OBJECTIVE To determine whether low-field MR fat/water separation and magnetisation transfer (MT) techniques are useful in studying the livers of patients with parenchymal liver diseases in vivo. METHODS MR and MT imaging of the liver in 33 patients (14 with primary biliary cirrhosis, 15 with alcohol-induced liver disease, and 4 with fatty liver) was performed by means of the fat/water separation technique at 0.1 T. The relaxation time T1 and the MT contrast (MTC) parameter of liver and spleen tissue were measured, and the relative proton density fat content N(%) and MTC of the liver were calculated from the separate fat and water images. The value of N(%) was also compared with the percentage of fatty hepatocytes at histology. RESULTS The relaxation rate R1 of liver measured from the magnitude image, and the difference in the value of MTC measured from the water image compared with the one measured from the fat and water magnitude image, both depended linearly on the value of N(%). The value of N(%) correlated significantly with the percentage of the fatty hepatocytes. In in vivo fatty tissue, fat infiltration increased both the observed relaxation rate R1 and the measured magnetisation ratio (the steady state magnetisation MS divided by the equilibrium magnetisation MO, MS/MO) and consequently decreased the MT efficiency measured in a magnitude MR image. The amount of liver fibrosis did not correlate with the value of MTC measured after fat separation. CONCLUSIONS Our results in studying fatty livers with MR imaging and the MT method show that the fat/water separation gives more reliable parametric results. Characterisation of liver cirrhosis by means of the MTC parameter is not reliable, even after fat separation.