Coal spontaneous combustion has always been a worldwide problem, which causes waste of coal resources, greenhouse gas emissions and other atmospheric environmental pollution problems. Although coal temperature monitoring is the most direct and accurate means of predicting the spontaneous combustion of coal, the coal temperature often cannot be directly measured owing to various physical restrictions. As an alternative, the present study assessed the qualitative and quantitative analysis of the CO and C2H4 formation rates, as well as various gas ratios such as the CO/CO2 ratio and fire coefficient R-2 (R-2 = 100x Delta CO/Delta O-2), to predict spontaneous combustion. This method was established based on the temperature-programmed experiments of three different coal rank (including lignite, bituminous coal and anthracite), and was verified using data obtained from on-site monitoring at an actual mine. The results show that the method accuracy is as high as 97% when predicting the coal temperature to within 15 degrees C (allowable error range of the predicted value). This degree of accuracy should be sufficient for on-site fire prevention and control. This new technique is not only accurate and reliable but also has theoretical significance with regard to the identification of coal spontaneous combustion in goaf and for the development of fire prevention and suppression technologies. (C) 2019 Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.