Diabetes mellitus is a chronic disorder, affecting nearly 200 mullion people worldwide. Acute complications, such as hypoglycemia, cardiovascular disease and retinal damage, may occur if the disease is not adequately controlled. As diabetes has no known cure, tight control of glucose levels is critical for the prevention of such complications. Given the necessity for regular monitoring of blood glucose, development of non-invasive glucose detection devices is essential to improve the quality of life in diabetic patients. The commercially available glucose sensors measure the interstitial fluid glucose by electrochemical detection. However, these sensors have severe limitations, primarily related to their invasive nature and lack of stability. This necessitates the development of a truly non-invasive glucose detection technique. NIR Raman Spectroscopy, which combines the substantial penetration depth of NIR light with the excellent chemical specificity of Raman spectroscopy, provides an excellent tool to meet the challenges involved. Additionally, it enables simultaneous determination of multiple blood analytes. Our laboratory has pioneered the use of Raman spectroscopy for blood analytes' detection in biological media. The preliminary success of our non-invasive glucose measurements both in vitro (such as in serum and blood) and in viva has provided the foundation for the development of feasible clinical systems. However, successful application of this technology still faces :a few hurdles, highlighted by the problems of tissue luminescence curd selection of appropriate reference concentration. In this article we explore possible avenues to overcome these challenges so that prospective prediction accuracy of blood analytes can be brought to clinically acceptable levels.