The human factor in the current railway system is a dominant cause of collision accidents. There exists a degree of uncertainty to the train driver on the timing of decision-making. Mitigating collision risks that are associated with the uncertainty of driving behavior is a challenging task. In this paper, we propose a train-simulator that incorporates the uncertainty of human driving behavior and its potential collision risk in the current train operation environment. The proposed simulator shows the impacts of driver behavioral errors on the safety performance and evaluates the collision risks associated with the various ranges of behavioral errors when a driver faces unsafe situations such as Signal Passed At Danger (SPAD) and signal failure events. The findings of this study reveal that there is a strong relationship between the number of taking evasive actions and the range of driving behavioral errors. This implies that the proposed train-simulator has a considerable effect in measuring the potential collision risks compared with the existing driving principles without the behavioral error consideration.