Abstract: It is in urgent need to solve the internal leakage of valves for it threatens the operation safety of natural gas pipelines. In this paper, a novel approach which combines wavelet packet decomposition and information entropy theory was developed to analyze the characteristics of acoustic emission signal of 3 in(1 in=25.4 mm) ball valve with different internal leakage levels. Firstly, the experiment platform used for the acoustic emission detection of valve internal leakage is built up to perform acoustic emission detection tests at different internal leakage rates. Then, the informativity of each frequency band of the acoustic emission signal is analyzed by means of the wavelet packet entropy method. And finally, the correlation between the wavelet packet entropy and the internal leakage rate in different frequency bands is evaluated by using root mean square (RMS) value as the evaluation parameter. It is indicated that the correlation between the wavelet packet entropy and the internal leakage rate is the highest (RMS error is 0.012 6) in the frequency band of 25-37.5 kHz. To sum up, this wavelet packet entropy analysis method is a new method to detect quantitatively the internal leakage rate of gas pipeline valves.