Abstract: This study has developed two-layer linear models using a method of progressively deleting insignificant explanatory variables based on the initial explanatory variables specified and by applying the principles of hierarchical linear model and quantitative theory, and analyzed the factors that influence the penetration rates of double-shield tunnel boring machines (TBM) using the in-situ data of tunnel No. 9 of the Tao River diversion project phase I. We also quantitatively describe the direct and indirect impacts by these factors, using model parameter estimates for the fixed effects of each explanatory variable in the models; based on these estimates, suggestions for better TBM operation are put forward. The results indicate that our method is feasible in prediction of the tunneling penetration rates, and it would be useful for forecasting the construction progress and control management in similar projects.