Abstract: The uranium ore heap in the stope of underground in-place leaching of fragmented uranium ore by blasting is a loose fragmented radon-emitting medium. Its particle size distribution subordinated to Rosin-Rammler distribution law. The medium is one of the main radon sources in the underground uranium mine, using in-place leaching of fragmented uranium ore by blasting. The transient radon emanation capability of this medium is a very important parameter for determining the amount of fresh air for ventilation and radon discharging in the stope of in-place leaching of a uranium mine and for simulating the radon emanation and transport in loose fragmented radon-emitting medium and is affected by transient radon concentration, characteristic particle size, particle size distribution index and moisture content of the medium.In order to study such influences, a self designed apparatus is used to make tests on seven samples which are compounded by using Rosin-Rammler distribution law and bared on the test results, the adaptive neuro-fuzzy inference system(ANFIS) is used to establish the ANFIS models for predicting transient radon emanation capability.The research results show that the transient radon emanation capability of loose fragmented radon-emitting medium increases with moisture content in the medium, and decreases with the moisture content when it increasese to a certain degree, and that the ANFIS models can give high accuracy predictions and this provides a new path for estimating transient radon emanation capability of loose fragmented radon-emitting medium.