H-NMR spectroscopy is a potentially useful technique for the characterisation of coal. A novel technique was applied to distinguish, characterise and determine quantitatively five different classes of proton containing components from air-dried Australian bituminous coals. The five different classes of protons are described by 14 fitted NMR parameters, which are the signal amplitudes, spin–lattice relaxation times and spin–spin relaxation times. To establish the mathematical relationship between the 14 fitted NMR parameters and the properties of a large set of bituminous Australia coals, a partial least-squares (PLS) regression algorithm was used. Predictions were accurate for many coal properties, particularly those related to the organic matter of the coal. The method is therefore, a rapid (25 min) method for coal quality monitoring with potential for commercial application.