A particularly challenging subject in the investigation of forensic human biological traces is analyzing samples containing mixtures of body fluids from multiple donors. Ideally, researchers want to identify each type of body fluid present. However, traditional methods, like mRNA and DNA profiling, often struggle with sensitivity, specificity, and efficiency, especially in complex mixtures. This proof-of-concept study has two primary aims: first, to classify body fluids within a mixture using discriminatory protein markers, and second, to evaluate the feasibility of using single amino acid variants (SAAVs) to trace the source of specific body fluids back to individual donors. To achieve this, we employed proteomic analysis via liquid chromatography-tandem mass spectrometry (LC-MS/MS) in data-independent acquisition (DIA) mode, developing a reliable approach for accurate body fluid classification. Through comprehensive proteomic profiling, we characterized a diverse array of discriminatory proteins present in peripheral blood, semen, saliva, urine, and vaginal fluid. Using advanced data analysis techniques, including t-distributed stochastic neighbor embedding (t-SNE), we demonstrated that these proteins could reliably distinguish between different body fluids, even in mixed samples. Additionally, our findings reveal that SAAVs within certain proteins, such as those in saliva, hold promise for source attribution in a forensic context. Challenges, including contamination and limited sample sizes, highlighted the need for strict quality controls and further large-scale studies. With these improvements, proteomic analysis could greatly enhance body fluid identification, classification, and source attribution in forensic investigations, improving both accuracy and reliability in forensic science.