The suppression of the abundant water signal has been traditionally employed to decrease the dynamic range of the NMR signal in proton MRS ( 1H MRS) in vivo. When using this approach, if the intent is to utilize the water signal as an internal reference for the absolute quantification of metabolites, additional measurements are required for the acquisition of the water signal. This can be prohibitively time-consuming and is not desired clinically. Additionally, traditional water suppression can lead to metabolite alterations. This can be overcome by performing quantitative 1H MRS without water suppression. However, the non-water-suppressed spectra suffer from gradient-induced frequency modulations, resulting in sidebands in the spectrum. Sidebands may overlap with the metabolites, which renders the spectral analysis and quantification problematic. In this paper, we performed absolute quantification of metabolites without water suppression. Sidebands were removed by utilizing the phase of an external reference signal of single resonance to observe the time-varying the static field fluctuations induced by gradient-vibration and deconvolving this phase contamination from the desired NMR signal. The quantification of metabolites was determined after sideband correction by calibrating the metabolite signal intensities against the recorded water signal. The method was evaluated by phantom and in vivo measurements in human brain. The maximum systematic error for the quantified metabolite concentrations was found to be 10.8%, showing the feasibility of the quantification after sideband correction.