Mastitis is one of the most impacting diseases in dairy farming, and its sensitive and specific detection is therefore of the greatest importance. The clinical evaluation of udder and mammary secretions is typically combined with the milk Somatic Cell Count (SCC) and often accompanied by its bacteriological culture to identify the causative microorganism. In a constant search for improvement, several non-enzymatic milk proteins, including milk amyloid A (M-SAA), haptoglobin (HP), cathelicidin (CATH), and lactoferrin (LF), have been investigated as alternative biomarkers of mastitis for their relationship with mammary gland inflammation, and immunoassay techniques have been developed for detection with varying degrees of success. To provide a general overview of their implementation in the different dairy species, we carried out a systematic review of the scientific literature using the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analysis) guidelines. Our review question falls within the type “Diagnostic test accuracy questions” and aims at answering the diagnostic question: “Which are the diagnostic performances of mastitis protein biomarkers investigated by immunoassays in ruminant milk?”. Based on 13 keywords combined into 42 searches, 523 manuscripts were extracted from three scientific databases. Of these, 33 passed the duplicate removal, title, abstract, and full-text screening for conformity to the review question and document type: 78.8% investigated cows, 12.1% sheep, 9.1% goats, and 6.1% buffaloes (some included more than one dairy species). The most frequently mentioned protein was M-SAA (48.5%), followed by HP (27.3%), CATH (24.2%) and LF (21.2%). However, the large amount of heterogeneity among studies in terms of animal selection criteria (45.5%), index test (87.9%), and standard reference test (27.3%) resulted in a collection of data not amenable to meta-analysis, a common finding illustrating how important it is for case definitions and other criteria to be standardized between studies. Therefore, results are presented according to the SWiM (Synthesis Without Meta-analysis) guidelines. We summarize the main findings reported in the 33 selected articles for the different markers and report their results in form of comparative tables including sample selection criteria, marker values, and diagnostic performances, where available. Finally, we report the study limitations and bias assessment findings.