This paper presents a new method (GA|ANN) developed by combining genetic algorithm (GA) and artificial neural networks (ANN) for determining parameters of soils and retaining walls of deep excavation. This method has the advantages of nonlinear projection of neural networks, networks reasoning, prediction and good overall characteristics. It was first used for back analysis of the problem of mechanics parameters for excavation. Case studies showed that the GA|ANN method is effective and practical for back analysis of determining parameters. This paper presents a new method (GA|ANN) developed by combining genetic algorithm (GA) and artificial neural networks (ANN) for determining parameters of soils and retaining walls of deep excavation. This method has the advantages of nonlinear projection of neural. Networks, networks reasoning, prediction and good overall characteristics. It was first used for back analysis of the problem of mechanics parameters for excavation. Case studies showed that the GA|ANN method was effective and practical for back analysis of determinant parameters.