The quantitative identification of water sources is an important prerequisite for objectively evaluating the degree of aquifer interference and predicting the production potential of coalbed methane (CBM) wells. However, this issue has not been solved yet, and water sources are far from being completely understood. Stable water isotopes are important carriers of water source information, which can be used to identify the water sources for CBM wells. Taking the Zhijin block in the Western Guizhou Province as an example, the produced water samples were collected from CBM wells. The relationships between the stable isotopic compositions of the produced water samples and the production data were quantitatively analyzed. The following main conclusions were obtained. (1) The δD and δ18O values of the produced water samples were between −73.37‱ and −27.56‱ (average −56.30‱) and between −11.04‱ and −5.93‱ (average −9.23‱), respectively. The water samples have D-drift characteristics, showing the dual properties of atmospheric precipitation genesis and water−rock interaction modification of the produced water. An index d was constructed to enable the quantitative characterization of the degree of D-drift of the produced water. (2) The stable isotopic compositions of produced water showed the control of the water sources on the CBM productivity. The probability of being susceptible to aquifer interference increased with the increasing span of the producing seam combination, reflected in the lowering δD and δ18O values and the decreasing gas productivity. (3) Three types of water, namely, static water, dynamic water, and mixed water, were identified. The characteristic values of the isotopic compositions of the static and dynamic water were determined. Accordingly, a quantitative identification method for the produced water sources was constructed, based on their stable isotopic compositions. The identification results have a clear correlation with the gas production, and the output of the static water contributes to the efficient CBM production. The method for the quantitative identification of the water sources proposed in this study, can help to improve the CBM development efficiency and optimize the drainage technology.