Abstract: A novel pre-search algorithm was proposed based on the spectral feature peak matching to address the existing problems of poor pre-retrieval precision and too many remaining spectra in mass spectral library search.In the newly-developed algorithm, the spectral peaks with the highest intensity and largest mass-to-charge ratio were defined as the characteristic peaks in the spectrum; then, each query spectrum was pre-retrieved by matching its characteristic peaks with those of all the reference spectra available.The results searched in both the novel and conventional algorithms were evaluated.The new algorithm outperforms TSLP algorithm with an accuracy improved by1.77%.The new algorithm works better than the VSM-based Mass Spectral Library Pre-search algorithm, not only because the accuracy was improved by 15.01%, 19.58% and 25.87%, respectively, with the thresholds of 0.6, 0.7 and 0.8, but also because the remaining spectra were significantly reduced.