Abstract: After compared with the classical Reich modal, improvements was made on an event-based aircraft collision risk model at intercepting routes, the collision risk due to possible vertical relative movement was counted for aircraft in straight and level flight. The possible reason why the horizontal collision probability derived by the event model was excessively conservative than the Reich model was analyzed. The aircraft relative position error data were generated using Monte-Carlo simulation and fitted to several probability distributions, their effects on aircraft collision risk were compared. The result shows that it is the tail data representing the gross error that decide the aircraft collision risk and no distribution functions except generalized Pareto distribution can fit the tail data well due to less amount of gross error data.