The seismo-acoustic method is one of the most promising emerging techniques for the detection of landmines. Numerous field tests have demonstrated that buried landmines manifest themselves at the surface through linear and nonlinear responses to acoustic/seismic excitation. The present paper describes modeling of the nonlinear response in the framework of the mass-spring model of the soil-mine system. The perturbation method used in the model allows for the derivation of an analytical solution describing both quadratic and cubic acoustic interactions at the soil-mine interface. This solution has been compared with actual field measurements to obtain nonlinear parameters of the buried mines. These parameters have been analyzed with respect to mine types and burial depths. It was found that the cubic nonlinearity could be a significant contributor to the nonlinear response. This effect has led to the development of a new intermodulation detection algorithm based on dual-frequency excitation. Both quadratic and intermodulation nonlinear algorithms were evaluated at the U.S. Army outdoor testing facilities. The algorithms appear to complement each other in improving the overall detection performance.