Forensic DNA analysis is well established for phenotyping, providing valuable investigative leads. Recently, proteomics, the large-scale study of proteins, is increasingly recognized as a complementary tool to DNA analysis, particularly for enhancing the evidential value of traces and especially in cases involving degraded samples or challenging matrices. This study aims to extract phenotypic traits directly from whole blood proteomes, using biological sex determination as a starting point. Using LC–MS/MS, proteomes from 100 whole blood samples of known sex were used to train a biological sex classifier. Cross-validation of the model demonstrated the potential of proteomics for accurate sex classification. Key peptides, such as from pregnancy zone protein and ceruloplasmin, were identified as highly important features. To further evaluate the model, mock case samples were generated to simulate real-world case scenarios. However, a large portion of these mock samples were incorrectly classified, which was caused by batch effects. Based on our findings, transitioning from an untargeted assay to a maximally performant analytical targeted assay is the next crucial step needed for implementation into routine forensic application. Overall, this study advocates for the inclusion of proteomics as part of the forensic phenotyping toolkit, while addressing the challenges, opportunities, and recommendations in its implementation.