Bend sensors fundamental characteristic is to furnish an electrical resistance value related to the angle they are bent. This feature can be successfully exploited to realize wearable systems capable to measure human static and dynamic postures. In particular some efforts have been made to determine finger joint movements of human hands and it has been demonstrated the feasibility of using the so called data glove system as a goniometric device. The repeatability of such system is quite good for general purposes but it is still not sufficient for specific applications (for instance in virtual surgery). So here we introduce a novel application method of bend sensors and demonstrate how it can be useful to improve the system repeatability.