A cornerstone of mass spectrometry based proteomics is to relate with high statistical significance experimentally obtained tandem mass spectrometry (MS/MS) data to peptide sequences from a protein database. Most sequence specific fragment ions in MS/MS spectra are represented by a subset of complementary ion pairs. Here, we investigated the reliabilities of complementary ion pairs formed in CAD and CAD/ETD MS/MS and developed a reliability-based approach of intensification of ion signals of complementary pairs prior to database searching. In a large-scale proteomics experiment using high-resolution orbitrap mass spectrometry, an increase in the number of peptide identifications was obtained relative to the original CAD MS/MS spectra when intensified golden complementary (+18.6%) and CAD complementary pairs (+17.2%) were submitted to the Mascot search engine. This also exceeded the results obtained by deisotoping/deconvolution of CAD MS/MS spectra. A novel approach for extracting sequence-specific fragment ions of co-isolated peptides was developed based on the complementarity rules. This technique demonstrated an impressive gain of 42.4% more peptide identifications as compared with the use of the initial data set.