Time-Domain Nuclear Magnetic Resonance (TD-NMR) was used to quantify the lipid contents of 48 different organic waste substrates. Results obtained from TD-NMR were compared to those from Soxhlet extraction, currently the prevalent method for organic waste characterization, especially in the field of anaerobic digestion. Two calibration methods were tested. The first was a self-calibration process using pure oils (NMR1) which showed good repeatability compared to Soxhlet extraction with a better coefficient of variation (5%). Analyses of volatile fatty acids (VFA) and long-chain fatty acids (LCFA) by chromatography were carried out to understand why the NMR1 method produced underestimations for some samples. Statistical analysis showed that the presence of saturated fatty acids had a significant effect on differences between the Soxhlet and NMR1 methods. The second calibration method applied chemometrics to TD-NMR raw data (NMR2), taking Soxhlet extraction values as references. It provided a good prediction of lipid content and avoided the lengthy calibration procedure usually required for this type of study. Last, the NMR2 method was shown to be highly suited to the quantification of lipids in organic waste, demonstrating better repeatability than the classic Soxhlet method.