Gravity and magnetic survey, the two most widely used and cheaply acquired geophysical datasets in ore deposit exploration, play an important role in the different stages of the mineral exploration. The recovered inverse models can be used to target regions of anomalous physical properties for further data acquisition or drilling. In recent years methods have been developed to calculate inverse solutions that predict the 3D distribution of physical properties required to explain the observed geophysical responses. The inverse calculation, deriving a physical property model from the data, is hampered by numerical problems and non uniqueness. Therefore，the non uniqueness of inversion demands that geological information be included. This paper started from model test, and provided a comparison of the inversion results constrained by different geological information. The model test revealed that, as more geological constraints are included, the inversion result is more similar with the true density model. A synthetic example of 3D gravity inversion using geological constraints in Lu Zong ore concentration district provided a 3D density model within 5km depth which shows the distribution of the main lithology volumes，the structures and the strata. It has proved in this thesis that the geological constrained 3D density mapping can enhance the resolution, optimize the inversion results and is very effective in making the ore concentration district perspective and prospecting the deep mineral.