The detection and quantitation of allergens as contaminants in foods using MS is challenging largely due to the requirement to detect proteins in complex, mixed, and often processed matrixes. Such methods necessarily rely on the use of proteotypic peptides as indicators of the presence and amount of allergenic foods. These peptides should represent the allergenic food in question in such a way that their use is both sensitive (no false-negatives) and specific (no false-positives). Choosing such peptides to represent food allergens is beset with issues, including, but not limited to, separated ingredients (e.g., casein and whey), extraction difficulties (particularly from thermally processed foods), and incomplete sequence information, as well as the more common issues associated with protein quantitation in biological samples. Here, we review the workflows that have been used to select peptide targets for food allergen detection. We describe the use and limitations of both in silico-based analyses and experimental methods relying on high-resolution MS. The variation in the way in which target selection is performed highlights a lack of standardization, even around the principles describing what the detection method should achieve. A lack of focus on the food matrixes to which the method will be applied is also apparent during the peptide target selection process. It is hoped that highlighting some of these issues will assist in the generation of MS-based allergen detection methods that will encourage uptake and use by the analytical community at large.