Abstract: The fast detection for nuclear materials according to the Coulomb scattering angle of the cosmic ray Muon is of great significance. Using the Muon scattering angle simulation data set of U, Pb and Fe obtained by Geant4, the distribution characteristics of Muon scattering detection data were analyzed. Testing the performance of the probability distribution function parameters and kurtosis of the Muon scattering detection data to classify different materials by support vector machine, and a fast detection algorithm based on distribution characteristics for nuclear material using cosmic ray Muon was proposed. The results show that the algorithm can only use 10 000 Muon scattering detection data to classify U, Pb, and Fe with the same thickness, and the classification accuracy rate is over 98.9%.