Iraqi western desert is characterized by a widespread karst phenomenon and caves. Euphrates formation (Lower Miocene) includes enormous sinkholes and cavities within carbonate rocks that usually cause severe damages to any kind of engineering facilities built over it. 3D resistivity imaging techniques were used in detecting this kind of cavities in complicated lithology. The 3D view was fulfilled by collating seven 2D imaging lines. The 2D imaging survey was carried out by Dipole-dipole array with (n) factor and electrode spacing (a) of 6 and 2m respectively. The horizontal slices of the 3D models give a good subsurface picture. There are many caves in all directions (x, y, z). They reveal many small caves near the surface. These caves are clearly shown as points with highly variable resistivity values in slices of depths 0.80 m, 1.72 m, 2.78 m, and 3.99 m. The comparison between standard Least-square and robust constrain methods appeared that the inverse model produced by the robust constrain method has sharper and straighter boundaries. The results of both the two-dimensional and the three-dimensional resistive imaging models deal with almost the same spread of subsurface caves in the study area and show a high amount in number, especially in the upper part, white in color oolitic limestone, the second member of Euphrates Formation,  of 3.8-8m depth. 2D and 3D resistivity imaging values have Standard Roof-mean square (RMS) error for large inverted models, this confirms that the study area is of high inhomogeneity. This heterogeneity resulted in a large variation in the resistivity of the rock component in addition to the large spread of caves near the surface.