Diabetes mellitus (DM) is a group of metabolic diseases related to the dysfunction of insulin, causing hyperglycaemia and life-threatening complications. Current early screening and diagnostic tests for DM are based on changes in glucose levels and autoantibody detection. This review evaluates recent studies on biomarker candidates in diagnosing type 1, type 2 and gestational DM based on omics classification, whilst highlighting the relationship of these biomarkers with the development of diabetes, diagnostic accuracy, challenges and future prospects. In addition, it also focuses on possible non-invasive biomarker candidates besides common blood biomarkers.