The monitoring of underground cavities plays a key role in risk management policies. Mine and underground quarry stakeholders require relevant methodologies and practices to define and assess hazards associated with these structures. To monitor these structures, geophysical methods may offer an interesting compromise among operating cost, invasiveness, and risk assessment reliability. The use of conventional 3D-electric resistivity imaging (ERI) software validated on relatively flat media is not sufficient to efficiently assess complex 3D geometries such as underground mine pillars. We have developed a new approach to evaluate pillar condition by means of a sequential use of two techniques. First, the photogrammetric method yields a detailed 3D model of the pillar geometry from a set of pictures. Second, 3D-ERI is performed based on this suitable geometry. The methodology is tested on a synthetic model to evaluate the effect of various geometry resolutions on the inversion. We also evaluated the combination of the effect of measurement and geometry error. We performed a quasi 3D-ERI survey (three parallel electrode lines) on a real limestone mine pillar to determine the benefits and limitations of the combined procedure. First results revealed the capacity of the photogrammetric methods to obtain a high-precision geometry and its key role during the inversion process. Second results of the real case study revealed that a highly accurate geometry is required todetect accurately conductive anomalies in a complex 3D context.