The recent emergence of the field of proteomics has been driven by advances in mass spectrometry methods and instrumentation. Due to the large amount of data generated, success at peptide and protein identification is contingent on reliable software algorithms. The software programs in use at the time the work in this dissertation was carried out were well suited to the task of identifying unmodified peptides and proteins in complex mixtures. However, the existing programs were not able to reliably identify protein modifications, especially unpredicted modifications. This dissertation describes the development of two novel software algorithms that can be used to screen LC-MS-MS data files, and identify MS-MS spectra that correspond to peptides with either predicted or unpredicted modifications. The first program, SALSA, is highly flexible and uses user defined search criteria to screen data files for spectra the exhibit fragmentation patterns diagnostic of specific modifications or peptide sequences. SALSA facilitates exhaustive searches, but requires user expertise to both generate search criteria and to validate matched spectra. The second program, P-Mod, provides automated searches for spectra corresponding to peptides in a search list. P-Mod is able to identify spectra derived from either modified or unmodified peptides. All sequence-to-spectrum matches reported in the P-Mod output are assigned statistical confidence levels derived using extreme value statistics.