One basic research for proteomics is protein identification. For large-scale protein identification, shotgun techniques are usually applied, i.e., some digested peptides (precursor ions) are chosen to fragment and produce tandem mass spectra, which can be identified by database search, and then proteins can be inferred from the identified peptides. In the identification, precursor mass is a key parameter. Whether the correct peptide is within the peptide candidates depends on whether the precursor mass is the monoisotopic mass. The number of peptide candidates depends on the accuracy of the precursor mass. This research investigates the accurate determination of precursors in terms of monoisotopic peak determination and systematic error elimination. There are some techniques of monoisotopic peak determination on protein level, including charge state determination, monoisotopic peak determination and overlapping cluster determination. Some of them can be used to the monoisotopic peak determination of precursors. Meanwhile, there are some well-known methods for systematic error elimination. The two kinds of techniques for accurate precursor determination can help increase the large-scale protein identification rate.