Background: Neuropeptides (NPs) play critical roles in synaptic signaling in various systems.NPs act as hormone,modulator,neurotransmitter and cytokine to regulate broad functions.NPs share the common characteristic that produced firom a longer NP precursor (NPP).With the drastic growth of unknown protein sequences generated in the post-genomic age,it is highly desired to develop computational methods for rapidly and effectively identifying NPPs.Method: Two predictors based on support vector machine,were contributed by dipeptide and tripeptide features respectively.Using the analysis of variance method,lower contribution features were ignored to make feature optimization in both predictors.Then the optimal attributes subset of features was used to develop the prediction model.Then the two predictors were integrated into a combined predictor to forecast NPPs robustly.