Human endogenous retroviruses (HERVs) comprise ~8% of the human genome. Despite traditionally being considered silent genomic relics, it is now clear that subsets of HERVs are transcriptionally active in human health and disease. Emerging evidence supports a role for HERVs in cancer, rheumatoid arthritis, systemic lupus erythematosus, and several neurodegenerative diseases, including motor neuron disease; however, direct proteomic evidence of HERV expression remains scarce. Detection of HERV-derived proteins is challenging due to their high genomic copy number, repetitive and redundant sequence features, and production via non-canonical translation mechanisms. Here, we explore computational strategies for the detection of HERV-derived peptides in mass-spectrometry-based proteomic datasets. Using curated sequence databases, we explicitly account for non-canonical open reading frames in both consensus and locus-specific HERV-K translations. These strategies are applied to examine the presence of HERV-derived peptides in publicly available proteomic datasets acquired from human biofluids. This work establishes a framework that will enable systematic interrogation of proteomic data for evidence of HERV expression and lays the foundation for exploring disease-associated HERV peptide signatures across the neurodegenerative disease spectrum and beyond.