The first requirement for an effective structural dynamic modification and control is an accurate finite element (FE) model having dynamic properties (natural frequencies and mode shapes, etc.) close to the measured values from real structures. The differences between the two models are attributed mainly to the uncertainties on joint regions, so an intermediate process to improve the FE model by adjusting and identifying joint properties is needed. A robust identification method is necessary to get reliable joint characteristics, and the selection of sensor positions plays a key role. All present methods of sensor positioning are a compromise between reliability and efficiency. This paper suggests a faster, automatic and more reliable sensor placement method. Based on an iterative scheme, it is similar to the effective independent method. Also, to overcome the shortcomings of the previous methods, it proposes a maximum allowable number of degrees of freedom that can be deleted at each iteration. This increases the ability to select optimum sensor positions. The suggested method was tested and verified with a satellite solar array and a T-shaped beam structure. (C) 1997 Academic Press Limited.