Researchers have been questioning if traffic microsimulation tools can be used for road safety evaluations. This thesis examines if these tools have the potential to predict conflicts between right-turning vehicles and through cyclists at signalized intersections. Moreover, this thesis evaluates if calibrating these models to describe the driving behaviour at signalized intersections significantly improves the conflicts’ prediction. It was found that VISSIM has the potential to predict traffic conflicts of interest. In particular, a moderate correlation was found between real conflicts and simulated conflicts of the default models (r=0.525). Calibrating the model for travel time improved the correlation between real conflicts and simulated conflicts (r=0.618). However, a one-way ANOVA test indicated that the improvement caused by travel time calibration was not significant. It was also found that VISSIM’s prediction accuracy is expected to decrease as either the cyclists’ volume or the product of cyclists’ volume and right-turning vehicles’ volume increase.