Abstract: Electrical Tomography (ET) is an advanced visualizing technique which is efficient, non-invasive and real-time, and has been widely used in industrial process detection and biomedicine diagnosis. As a priori information and an indispensable part, the sensitivity coefficient plays a significant role in the ET process. However, the existing calculation method of sensitivity coefficient has the drawbacks of ambiguity, lack of explanation capacity, poor adaptability and linear hypothesis, which results in an obstacle of improving the quality of reconstructed images and restricts the development of ET process. This paper starts from the physical significance of sensitivity coefficient in quasi-static field. In the Electrical Resistance Tomography (ERT) field, the relationship between each pair of excitation electrodes can be equivalent to a pair of dipoles. Furthermore, each corresponding measured point can be equivalent to a pair of dipoles. Then, corresponding to the distribution of physical field, a new method of calculating the sensitivity coefficient is proposed, which has the advantages of clear physical significance, strong interpretability and operability. Finally, in the COMSOL with MATLAB platform, a set of experiments was established to test the performance of the new method. During the simulation, the new sensitivity coefficient was compared with the original sensitivity coefficient in the numerical, projection area and calculation time. Besides, the two different sensitivity coefficients and the sensitivity coefficient back projection (BP) method were used to reconstruct the four typical models: three circles model, discrete vesicle model, double lung model and cross model. Through a series of simulation and analysis, the results show that the new method is feasible and consistent with the numerical distribution of the original calculating results. Moreover, the new sensitivity coefficient solution has better performances in noise immunity, computation time and image reconstruction.