A database of tunneling parameters and strata was established considering the shield tunneling practice of Guangzhou Rail Transit. Based on the data, a method of strata identification was studied by using neural network pattern recognition technology. Based on the analysis of the features of strata in shield tunneling and the data, a one-to-many mapping relation between strata and data was proposed, as well, the strata identification not only to contrast between the parameters, but also to contrast between the combined effects of each parameter mapping was pointed out. On the basis of what was mentioned above, a three-layer BP neural network model was built. What is more, some typical tunneling process parameters were input as training data, and the test results had a good agreement with practical situation. The test result could be used as strata identification. This method will enhance the scientificalness of shield tunneling control, and the timeliness and speediness of it are helpful to automatic driving for shield machine.