Abstract: Coal spontaneous combustion is the main disaster affecting mine safety, and the indicator gas is one of the most effective parameters for early warning of coal spontaneous combustion.In order to grasp distribution characteristics of gas concentration in goaf and to provide data support for early warning of coal spontaneous combustion, the distribution characteristics of gases in the goaf are analyzed on the basis of statistical theory, and the periodicity of CO concentration change in the goaf is analyzed by using wavelet transformation.The results show that the O₂ concentration obeys the normal distribution, and the CO concentration obeys the log normal distribution, and the coal spontaneous combustion "three belt" distribution in the same flat area has similarity, and the width of air leakage band in air ward is about 10 m, and the width of wind side leakage is greater than the return side, the width of oxide band is about 20-30 m longer compared to the leakage bandwidth, the CO concentration in goaf has significant periodicity, the first main cycles and second main cycles exist, and in the first main cycle, the change of CO concentration conforms to the distribution characteristics of "high-low-high".The distribution period is the same as the working surface periodic pressure time.The results are of great engineering significance to master the gas distribution law and coal spontaneous combustion warning in the goaf.