Abstarct: The magnetotelluric (MT) method is a non-invasive electromagnetic geophysical technique that utilizes natural variations in the Earth's electric and magnetic fields over a wide frequency range to image the subsurface electrical resistivity distribution at varying depths. Due to its high sensitivity to conductive structures and its capability for deep penetration, MT is considered an effective tool for identifying and modeling salt diapirs. Dry salt typically exhibits high electrical resistivity and strong contrast with surrounding lithologies, making it readily distinguishable in MT data. As part of a natural gas storage feasibility study in the Nasrabad region of Kashan, various geophysical surveys including magnetotellurics and seismic reflection were conducted over the area's salt diapirs. Seismic methods failed to accurately resolve the geometry and depth of salt diapirs No. 4 and 5. To supplement this limitation, MT data were acquired at 253 stations along six profiles. Analysis of the apparent resistivity and phase derived from impedance tensor data revealed significantly lower resistivity in diagonal components compared to off-diagonal ones across all stations, indicating a dominantly two-dimensional (2D) structural regime. Dimensionality analysis using the phase tensor confirmed a nearly 2D environment at periods less than 10 seconds, transitioning to 2D/3D behavior at longer periods. The regional geoelectric strike direction, estimated using phase tensor analysis, was approximately N30°W, and data were rotated accordingly for 2D and 3D inversion. Vertical magnetic field components were also recorded at select stations, enabling tipper vector analysis. This analysis indicated the presence of conductive structures in the northeastern sector and more resistive features in the southwestern part of the study area. Electromagnetic skin depth estimation using the Niblett–Bostick depth approximation showed penetration depths exceeding 20 km at all stations. Two-dimensional inversions using TE, TM, combined TE+TM, and determinant modes were performed along all profiles. The resulting models successfully imaged subsurface structures, including salt diapirs No. 4 and 5, and delineated the geological laxyering both laterally and vertically. However, due to the inherently complex and three-dimensional nature of the salt diapirs, three-dimensional (3D) modeling was necessary for more accurate geometric characterization. A total of 21 isotropic 3D inversion tests were conducted using various data configurations, initial resistivity models, and covariance values. The most favorable isotropic 3D model was achieved using full impedance tensor data and a covariance factor of 0.6. Additionally, anisotropic 3D inversions were carried out in eight tests using full impedance tensor data, both rotated and unrotated relative to the strike direction, with varying covariance values. The optimal anisotropic model corresponded to rotated data with a covariance value of 0.7. The final anisotropic 3D inversion model effectively resolved the geometry of diapirs No. 4 and 5, particularly the positions of their tops and baxses. The dimensions of diapir No. 4 were estimated at approximately 2 km × 1.5 km, and diapir No. 5 at approximately 2 km ×1 km. By contrast, the 2D and isotropic 3D models only provided preliminary images of these structures along profiles 21 and 23. Comparison of the MT inversion models with well log data showed a reasonable level of agreement.