A method and apparatus for noninvasive glucose measurement measures glucose indirectly from the natural response of tissue to variations in analyte concentration. The indirect measurement method utilizes factors affected by or correlated with the concentration of glucose, such as refractive index, electrolyte distribution or tissue scattering. Measurement reliability is greatly improved by stabilizing optical properties of the tissue at the measurement site, thus blood perfusion rates at the sample site are regulated. Perfusion is monitored and stabilized by spectroscopically measuring a control parameter, such as skin temperature, that directly affects perfusion. The control parameter is maintained in a range about a set point, thus stabilizing perfusion. Skin temperature is controlled using a variety of means, including the use of active heating and cooling elements, passive devices, such as thermal wraps, and through the use of a heated coupling medium having favorable heat transfer properties.