NMR methodology is developed for high-resolution, accurate measurements of methyl H-1(m)-C-13(m) (D-1(CH)) and C-13(m)-C-13 (1 D-CC) residual dipolar couplings (RDCs) in ILV-methyl-protonated high-molecular-weight proteins. Both types of RDCs are measured in a three-dimensional (3D) mode that allows dispersion of correlations to the third (C-13(beta/gamma)) dimension, alleviating the problem of overlap of methyl resonances in highly complex and methyl-abundant protein structures. The methodology is applied to selectively ILV-protonated 82-kDa monomeric enzyme malate synthase G (MSG) that contains 273 ILV methyl groups with substantial overlap of methyl resonances in 2D methyl H-1-C-13 correlation maps. A good agreement is observed between the measured RDCs of both types and those calculated from the crystallographic coordinates of MSG for the residues with low-amplitude internal dynamics. Although the measurement of D-1(CH) RDCs from the acquisition dimension of NMR spectra imposes certain limitations on the accuracy of obtained D-1(CH) values, D-1(CH) couplings can be approximately corrected for cross-correlated relaxation effects. The ratios of D-1(CH) and D-1(CC) couplings (D-1(CH)/D-1(CC)) are independent of methyl axis dynamics and the details of residual alignment [Ottiger, M.; Bax, A. J. Am. Chem. Soc. 1999, 121, 4690.]. The D-1(CH)/D-1(CC) ratios obtained in MSG can therefore validate the employed correction scheme.