Estimating the field penetration index (FPI) is an essential task in tunneling as the FPI is used to determine the tunnel boring machine (TBM) performance. In this study, fuzzy inference system (FIS) modelling is implemented to predict the FPI. Several models including fuzzy clustering and knowledge-based models were proposed. Data from the Queens Water Tunnel underneath Brooklyn and Queens were used to establish and validate the models. The input parameters include the rock type, uniaxial compressive strength, Brazilian tensile strength, rock brittleness (BI) of the intact rock, the angle between the plane of weakness and the TBM driven direction (Alpha), the distance between planes of weakness (FS), and the TBM cutter load. In order to evaluate the effect of the characteristics of the fractures on the FPI prediction, several models with different inputs and dataset structures were explored. The models were tested with independent datasets and performance indices used included the coefficient of determination R2, values account for (VAF), root-mean square error (RMSE) and mean absolute percentage error (MAPE). Overall, the model performance results were satisfactory with R2, VAF, RMSE and MAPE ranging between 0.79–0.92; 79.42–92.06%; 6.66–11.05; and 5.68–8.96%, respectively indicating good predictability capability. The models based on fuzzy clustering yielded higher accuracy. It was established that BI, Alpha and CL were the parameters controlling mostly the FPI. Based on that, the knowledge-based model was developed and satisfactory results were achieved as well. It was concluded that the FISs could be used to estimate the FPI values with a reliable accuracy.