Abstarct: Nowadays attempts to detect and achieve buried structures and underground resources have developed widely and geophysics is a means of identifying these structures. The main goal of geophysical methods, including gravimetery, is ncorporating additional information to the pross of inversion to define the characteristics of geological structures as precisely as possible. Since our country is rich in mineral resources, using modern and efficient methods to optimize and prevent from additional costs is very important. So in the first step; i e, exploration, we should try to collect as much comprehensive information about the underground structure as possible so that in the next costly step, that is extraction, all the necessary policies are chosen and the expenses are minimized. To reach this goal 2D and 3D modeling is one of the most successful methods of interpreting geophysical data. The outcome of modeling is to have a better understanding of target such as the shape and depth of source. Finding these parameters influences the subsequent decisions directly which can influence the management of major costs. The aim of this study is 3D constrained inversion of gravity data to detect subsurface structures with Li-Oldenburg method (GRAV3D software). The two main objectives of this method are to identify the shape of source and determination of density contrast of the anomaly with the background. This algorithm includes a multi-coefficient objective function which is flexible enough to produce different kinds of models which are in good agreement with geological structures. Incorporating additional information to the process of inversion is the strength of this algorithm which can be done with different constraints such as smoothness, positivity, reference model and bounded model. Constrained modeling leads inversion towards the production of logical models and as a result, the validity and reliability of the final model will increase. For optimized use of the algorithm at first it was tested on artificial models which were contaminated with noise, then according to the accommodation of the results with the right model it was used for the inversion of real data. In this research, gravity data of Sabzevar chromite exploratory area, Qom salt dome and Humble salt dome (located in the U.S.A) were inverted and the final results which have been shown in different views.