The authors present a method for detection of conflicts in airspace within look-ahead time of 20 minutes, based on analysis of the current flight plans. It combines both geometric and probabilistic approaches, and contains, respectively, geometric and probabilistic algorithms. The geometric conflict detection algorithm forms the basis of the method. It views an aircraft trajectory as a series of straight line segments between the route points. In addition, it handles the dynamic evolution of the trajectory uncertainty with increasing look-ahead time. This algorithm ensures minimization of the number of conflict misses due to estimation of the distance between almost entire areas of predicted aircraft positions. The algorithm for calculation of conflict probability forms a superstructure of the method. It estimates the probability level of an event, when the distance between the two conflict aircraft is less than the threshold value at a closest approach time. This algorithm produces a supplementary indicator of a situation criticality. Two algorithms of probabilistic kind were developed and tested. Their results differ only in case when the predicted minimal approach distance is approximately equal to the threshold value. The algorithm, which gives greater value to the probability level, is chosen. The method is fully analytical, so it does not need step-by-step simulation. The appropriate computer program is used as a part of the Air Traffic Control system in the airport of Minsk, Belarus, providing to the controllers a considerable amount of information on the predicted air situations.