Abstract: Open search can identify all modifications of masses within a given wide precursor mass tolerance window. But open search does not account for modified fragment ions when matching peptide with spectrum, resulting in reduced sensitivity compared to traditional variable modification searching. Inserting complementary ions is an intuitive strategy to address this weakness. In this study, different spectrum processing methods making use of complementary ions were investigated. DeconTop20, which deisotope and deconvolute the spectrum, then adding the top 20 peaks’ complementary ions, was shown to be the best method for open search. The identification of peptides for open search relies only on the match between peptide fragment ions and MS/MS spectrum peaks, the filter power of precursor mass is ignored, which makes open search less sensitive than close search and more dependent on the resolution of tandem mass spectrum. Even using DeconTop20, the identification sensitivity of open search decreases rapidly when the resolution of mass spectrometry decreased. Clearly, the conventional complementary ion based method is not efficient to improve the identification sensitivity for low resolution data. Complementary ion spectrum is the spectrum composed of all the complementary ions of its original spectrum, which contains equivalent peptide-fragment information. The probability for a spectrum and its complementary ion spectrum matched to the same decoy peptide sequence is much lower than they matched to the same target peptide sequence. Based on this principle, a complementary ion spectrum strategy was proposed to improve the sensitivity of the open search. In this strategy, only the identical peptides identified by both the complementary ion spectrum and its original spectrum were considered as a valid match. In this study, the average ratio for a spectrum and its complementary ion spectrum match to the same decoy peptide sequence was 15%, which meant 85% false positives matches could be filtered out with this constraint, and a great increase of identification number were achieved under given false discovery rate. The sensitivity of this strategy increases significantly compared to conventional open search as the fragment ion tolerance increases. When applying 0.05 u fragment ion tolerance, the numbers of peptide-spectrum matches and peptides with 1% FDR increased by 40% and 42%, respectively; if 0.5 u fragment ion tolerance was used, the increments get to 116% and 97%. Clearly, this strategy can greatly reduce the dependence of open search on the resolution of mass spectrometry.