Abstract：Stochastic conjugate gradient inversion，a new nonlinear inversion technique，is presented here. It works in the following steps:Nonheuristic inversion method is taken to fast converge to a local extremum.Heuristic inversion method is taken to jump out the local extremum.Nonheuristic inversion method is used again to converge to another local extremum;and the above process is repeated until a desirable situation appears.Search in solution space is done to save all local extremums，from which the optimum solution can be derived finally.The technique finds optimum solution rapidly because it inherits both global search of stochastic hillclimbing method and fast accurate computation of conjugate gradient method. It has been shown that the technique is high efficient and particularly applies to nonlinear multiextremum optimization problem and geophysical inverse problem.
