The application of Iterative Discriminant Analysis (IDA) and Neural Networks (NN) methods for e

μ

π particle identification in BES experiment on BEPC collider was discussed. Under exactly the same condition

satisfactory identification efficiencies and background levels were given by both methods for those particles with momentum in the range from 0.2GeV to 1.6GeV. Although the momentum spectrum of the data samples used for training were non-uniform

it was interesting that the selection efficiencies given by test results from both method were quite uniform in the whole momentum range. It shows that the NN method is slightly better than IDA method. Their characteristics and application condition were also discussed.