Abstarct: In recent years, TBMs are widely used in tunneling projects all over the world. Since the mechanized tunneling started a wide variety of performance prediction models have been developed. Most of the prediction models have considered only penetration rate. Some models have been developed for prediction of TBM advance rate and have suggested some relationships for estimating utilization factor. In TBM tunneling projects, prediction of the advance rate is very important as it has a big influence on the duration of the project and the costs and utilization factor also has a great effect in advance rate. Different Rock mass conditions have a great role in on TBM downtimes and each effect may reduce machine utilization. The main goal of this research is the study of effect of rock mass condition on utilization factor. In order to model the effect of geology and rock mass conditions on TBM downtimes (GRRD), a thorough databaxse baxsed on three long water conveyance tunnels in Iran has developed. Developed databaxse contains maps, detailed engineering geology information and daily site reports including length of excavated tunnel in each rock mass units, daily boring time and advance and different rock mass related downtimes. To achieve this, detailed maps, engineering geology reports and the daily site reports of study cases including Karaj-Tehran water conveyance tunnel (lot-1 length: 16 km and lot-2 length: 11 km of 14 km) and Ghomrood water conveyance tunnel (lots 3 and 4 length: 18 km). This effect has been modeled by using regression analysis, combination of Rock Engineering Systems (RES)-regression analysis and Fuzzy logic methods. Various regression equations were developed between GRRD and RMR and some meaningful significant relationships were achieved. After statistical analysis, the quadratic equation showed the correlation between RMR and GRRD more accurately with the coefficient of determination 0.62. Using RES method the geology and rock mass related downtimes index (GRDi) were calculated and then regression analysis was performed. In this approach, GRDi values are independent input variables whereas the measured GRRD is a dependent variable. The maximum coefficient of determination (R2 = 0.85) was obtained in quadratic equation. Also Fuzzy logic method is capable to model GRRD with a good coefficient of determination (R2 = 0.97). Comparing these 3 models shows that regression analysis is easier to use but RES-regression model is capable to consider more effective qualitative parameters such as poisonous gases, mixed face condition, existing clay, squeezing and rock abrasion.