For more than thirty years there has been a high demand for glucose measuring devices to replace the finger pricking methods and to improve the life quality of diabetics. Three approaches are possible, electrochemical, electromagnetic (including optical), and mechanical. We analyzes the feasibility of a glucose sensing technology based on the rheological property of a liquid solution containing Concanavalin A, dextran and glucose which changes its viscosity with the glucose concentration. This sensitive solution is placed into a minimally-invasive subcutaneously implantable sensor. The glucose equilibrium establishes through a selective membrane. The viscosity is measured with a passive, magnetically actuated two-rotor microviscometer. The first rotor measures the local temperature and the second the glucose dependent viscosity of the solution. To miniaturize the microviscometer, a sensitive solution with a low viscosity (<30mPas at 37°C) is required. Different compositions, between 3-5% dextran 2000 and 0.5% and 0.8% Concanavalin A have been prepared and their viscosity as function of the glucose concentration (2-30mM) has been measured between 20 and 40°C. The measurements were performed in a thermostatic chamber with a glass capillary using a camera and an image recognition software. Over this physiological glucose range, relative viscosity variations of up to 1000% were obtained. The passive contact-less microviscometer is described in a one- and two-rotor configuration. The accuracy of the one-rotor system is better than 1% in the range of 1.35 to 17.25mPas. The accuracies for the two-rotor version, in the range 20 to 40°C, is better than ± 0.2°C for the drive capsule (used as thermometer) and better than 3% for the active capsule (for glucose sensing) in the range 4.6 to 17.3mPas. Both versions have first been characterized in detail with silicone calibration oils. To show the feasibility of the concept the two-rotor system was then evaluated in real in vitro glucose measurements with a flat membrane. A simplified interstitial fluid with variable glucose concentration simulates the in vivo conditions. The semipermeable membrane used as interface between the sensing liquid and the interstitial fluid is based on anodic alumina membranes with 20nm pore size. A Polyethylene glycol (PEG) coating narrows the pore size to retain dextran and Concanavalin A but still allows glucose exchange. In addition this coating enhances the biocompatibility of the membrane and was designed to hinder an adverse immune response or a severe fibrous encapsulation. The reversibility and the fast response time of the glucose affinity sensing technique have been demonstrated and the potential accuracy of this method is shown in several measurements. The first prototypes use a rather large amount of sensitive solution and therefore show a lag time of 20 to 30 minutes. Solutions to decrease this time are discussed.