A key challenge in the analytical assessment of therapeutic proteins is the comprehensive characterization of their higher-order structure (HOS). To directly assess HOS, a new type of assay is warranted. The most sensitive and detailed method for characterizing HOS is unquestionably nuclear magnetic resonance (NMR) spectroscopy. NMR spectroscopy provides direct information about the HOS at an atomic level, and with modern NMR spectrometers and improved pulse sequences, this has become feasible even on unlabeled proteins. Hence, NMR spectroscopy could be a very powerful tool for control of HOS following, for example, process changes resulting in structural changes, oxidation, degradation, or chemical modifications. We present a method for characterizing the HOS of therapeutic proteins by monitoring their methyl groups using 2D H, C-correlated NMR. We use a statistical model that compares the NMR spectrum of a given sample to a reference and results in one output value describing how similar the HOS of the samples are. This makes the overall result easy to interpret even for non-NMR experts. We show that the method is applicable to proteins of varying size and complexity (here up to ∼30 kDa) and that it is sufficiently sensitive for the detection of small changes in both primary and HOS.