Previously, the theoretical relationship between paramagnetic chemical shifts and the axial ligands in low-spin haem proteins has been tested extensively in haems b and c with His, Met, and cyanide ligands. Variations in the electronic structure of the haem and the magnetic susceptibility tensors have been shown to depend primarily on the axial ligand geometry, and the shifts of haem substituents have been used to obtain the first structural information for several cytochromes. Recently, the database of assigned spectra for bis-His haems has been extended sufficiently for an empirical equation to be produced for treating 1H NMR data from haem methyl groups at 298 K. However, the database used contains large systematic deviations and the form of the equation leads to systematic errors in the ligand geometries. This article describes the link with the semi-empirical methods used previously and provides a set of corrected empirical parameters as well as an improved equation. The possibilities for generalising the empirical method to account for ligands other than His and temperatures other than 298 K are discussed.