Recent improvements in sensitivity have enabled direct binding studies of small molecules with evanescent wave biosensors, which monitor binding by measuring refractive index changes close to the sensing surface. The universal solvent for small molecules, dimethylsulfoxide has a high refractive index; consequently, on ligate addition a large non-specific solvent effect is seen which can mask the specific signal. It has been previously noted that different sensor surfaces can respond differently to the same buffer change. The difference is proposed to arise from differences in buffer space and contraction and swelling of the surface hydrogel. Within this paper, a number of calibration approaches are investigated and tested using warfarin binding to human serum albumin as a model system. A number of recommendations are made for accurate referencing for non-specific effects. Changes to the ionic strength of the running buffer had little effect, whilst changes to the charge density of the carboxylmethyl dextran significantly affected how well the control surface reflects the non-specific signal. An amended 'calibration method' can be used, however, it is an additional complex step that was found to overcorrect in the presence of non-specific binding. Matching immobilisation levels between control and active surface significantly reduces solvent differences allowing accurate correction providing solvent compositional changes are minimised in experimental design. Under these circumstances, the traditional method of simple subtraction of the control from the active response is the most appropriate method of correction.