Abstract Hepatic phosphorus-31 magnetic resonance spectroscopy ( 31P MRS) was undertaken in 85 patients with histologically proven cirrhosis of varying etiologies and functional severity. Reference data were acquired from 16 healthy volunteers who had no history or evidence of liver disease or alcohol abuse. In vivo hepatic 31PMR spectra were acquired with pulse angle 45° and repetition times (TR) of 5 and 0.5 seconds. Peak area ratios of phosphomonoesters (PME), inorganic phosphate (Pi), and phosphodiesters (PDE) relative to βATP, and of PME relative to PDE were calculated from spectra acquired at TR 5 seconds. Estimates of saturation effects for individual resonances were obtained by dividing the peak height at TR 5 seconds by that at TR 0.5 seconds to yield a T I-related signal height ratio (SHR). When compared with reference values, the patients with liver disease showed a significantly higher PME/ATP ( P< .0001), PME/PDE ( P < .0001), PME SHR ( P < .001), and Pi SHR ( P < .02), and a lower PDE/ATP ( P < .001) and PDE SHR ( P < .001). The magnitude of these changes increased significantly and progressively with increasing functional impairment. In patients with compensated cirrhosis spectral appearances varied with etiology; thus, patients with postviral cirrhosis showed a significantly higher Pi/ATP; those with alcoholic cirrhosis, a significantly lower PDE/ATP; and those with cirrhosis secondary to primary sclerosing cholangitis, a significantly lower Pi/ATP than the healthy volunteers or other etiological groups. However, spectral appearances did not vary with etiology in patients with decompensated disease. In vitro 31P MRS of perchloric extracts of samples of liver tissue obtained from 10 patients with cirrhosis at transplant hepatectomy showed increases in levels of the soluble PME metabolites, phosphorylcholine and phosphorylethanolamine, and reductions in the levels of the soluble PDE metabolites, glycerophosphorylcholine and glycerophosphorylethanolamine. These changes suggest regenerative activity in cirrhotic livers. The increases in soluble phosphomonoesters in the aqueous extracts accounted for the increased PME/ATP ratio seen in the in vivo spectra, and might account for the increase in PME SHR. The reduction in soluble phosphodiesters in the aqueous extracts did not entirely account for the reduction PDE/ATP ratio seen in vivo. This change in the PDE resonance in hepatic in vivo spectra may also be partly due to reduced contribution to this peak from hepatic endoplasmic reticulum, resulting from replacement of hepatocytes by fibrous tissue. Hepatic 31P MRS provides insights into the pathophysiology of liver injury and may contribute to the assessment of hepatic functional state.