¹H-MR spectra of the liver contain spectral lines of glycogen (Glycg) and choline-containing compounds (CCC) in addition to the strong water and lipid (Lip) resonances. Quantification of CCC is important because choline is related to malignancy.^1,2 Liver glycogen serves as a form of energy storage. It can be quickly mobilized to meet a need for glucose.³ Glycogen and lipid metabolism is involved in metabolic disorders (diabetes mellitus, obesity).^3-5 However, quantification of Glycg, CCC and their relaxation times is a challenge. Small Glycg and CCC spectral lines “disappear” from the spectrum already due to slight respiratory motion. In the present work, we applied multiple breath-hold ¹H-MRS technique to obtain liver spectra. Our aim was to measure Glycg, CCC, water, and lipid (-CH₂-)_n relaxation times T₁, T₂ and to estimate absolute concentration of liver Glycg, CCC and fat.Two groups of healthy volunteers were recruited. The first group (5 males, median age 26 years, range 24-64) participated in relaxation time experiments. Metabolite concentrations were estimated using the second group (8 males, median age 41.5 years, range: 28-64, mean body mass index 26.2±1.2 kg/m², range: 24.3-27.9). Experiments were performed on a Philips (Achieva) and General Electric PET/MR (Signa) 3T scanners. Expiration breath-hold (EBH) single-voxel PRESS sequence was used for localization (Fig. 1). Relaxation time experiments were performed on a Philips scanner (BW 2000 Hz, voxel size 35x35x35 mm³). Water T₁, T₂ were computed using the spectra acquired at seven TRs (range: 710-4000 ms, two dummy excitations, number of scans (NS) 2, TE 40 ms), and seven TEs (range: 40-175 ms, TR 1500 ms, NS 2). Water suppressed (WS) spectra were used for estimation of Glycg, CCC and lipid relaxation times. Three spectra were acquired for each combination TR and TE. Two dummy excitations were followed by 4 scans. T₁ values were estimated from the spectra measured at TE 60 ms and TRs between 720, and 300 ms. T₂ values were computed from the spectra measured at TE range 40-150 ms (TR 1500 ms). Summed signals of Glycg H2H4’, H3 and H5 resonances (Fig. 2) were used to estimate the apparent relaxation times.⁵ Glycg, CCC and lipid content were quantified using the spectra collected on a GE scanner: voxel size 30x30x30 mm³, BW 5000 Hz, 4096 points, TR/TE 2000/35 ms. 16 non-WS scans were followed by 32 WS acquisitions. FIDs were acquired in a “two-scans mode” (2 phase cycle steps, NS 2), which means that 8 non-WS and 16 WS FIDs were saved separately and processed by LCModel. Relaxation times were estimated by monoexponential fitting of the spectral intensities (Fig. 3). Lipid, CCC and Glycg content was estimated from relaxation corrected spectral intensity ratios to unsuppressed water line. Reference concentration 37 320 mM of “NMR-visible” water in the liver was used. We assumed that liver contains 0.711 g H₂O per 1 g wet weight tissue and its density is 1.051 g/cm³.^4,5 Furthermore, it was assumed that 10% of tissue water is “NMR-invisible”. Lipid concentration was quantified using (-CH₂-)_n intensities.⁶ Glycg concentration was estimated from the spectral intensities H2H4’, H3 and H5.⁵ Despite of breath-hold were many spectra distorted by involuntary movement. Corrupted spectra were excluded. Larger voxels were used to improve SNR. Figure 2 shows representative WS spectrum of healthy liver. Relaxation times and metabolite concentrations are shown in Tab. 1. To our knowledge, this is the first 3 T MRS study in which Glycg, CCC, and lipid T₁ relaxation times of healthy human liver were estimated. T₂ relaxation times of Glycg and CCC and their concentrations can only be compared with the results of Ouwerkerk et al.⁵ Our values (Tab. 1) are in good agreement with this study. Multiple breath-hold ¹H-MRS technique improves quantification of liver Glycg and CCC.