Under the background of a computer wargame system, a trajectory clustering algorithm named CTECW (clustering trajectories of entities in computer wargames) is proposed. The algorithm is composed of three parts: trajectory pretreatment, trajectory segments clustering, and visual presentation. Trajectory pretreatment transforms original trajectories into simplified ones which are ulteriorly processed into linear segments. In the second part, the concept of density function derived from DENCLUE is introduced and trajectory segments are clustered based on similarity measure under the framework of DBSCAN. The visual presentation exhibits clusters of trajectory segments with martial meanings to trainees, which embodies practical values of CTECW. Both theoretical analysis and experimental results indicate that CTECW could acquire approximate clusters more efficiently compared with TRACLUS and requires no input parameters.