The risk of spontaneous combustion of coal is highly serious especially in gaseous underground coal mines. In many cases such a spontaneous combustion is a source of initiation of methane-explosive mixture with tragic consequences. Early indication of spontaneous combustion and determination of its seat temperature is in a given environment a key part of safety of underground coal mines. A commonly used method for the detection of spontaneous combustion is an interpretation of coal oxidation gas products whose historical roots go back to the first halve of the 20th century as Graham's ratio. In the second half of this century a wide research on application of desorbed higher hydrocarbons C1-C4 and hydrogen for the given purpose took place. Until now, the interpretations in question encounter significant inaccuracies. The paper describes the research on factors affecting the accuracy of the determination of coal spontaneous combustion temperature on conditions of underground coal mines. The research was focused on the accuracy of interpretation of mine gases in connection with the use of laboratory verified gas characteristics of thermal oxidation of coal. Three chromatograph laboratories had been involved to test thermal oxidation of coal to determine a deviation of spontaneous combustion gases. The variability among the laboratories and used techniques were insignificant. Besides the laboratory methods, it may not reflect processes in coal mines and one of the results was that coal itself had been a source of substantial variability of desorption of spontaneous combustion gases. The spontaneous combustion data still requires human interpretation. Simple criteria for providing warning by a computer do not exist just yet. (C) 2010 Elsevier B.V. All rights reserved.