Abstract: The development of oil and gas in ultra-deep reservoirs has grown by leaps and bounds. The predication of rock mechanics in ultra-deep formations, however, is a challenge because the geological structure is complicated, and cracks are huge so that the accuracy based on traditional methodology is difficult to accept. Hence, this paper developed a predication model to predicate rock mechanics in the whole ultra-deep reservoir. This model, based on laboratory data and on-site field data, was a multivariate nonlinear regression fitting model which was constrained of the relationship between logging respond, natural fracture conditions, and rock mechanics properties. Comparing with the achieved date after actual drilling and fracturing work in the field, the accuracy of the results in the model can be up to 90%. This predication model provides reliable rock mechanical data to support for field drilling and completion work in ultra-deep formations.