Recognition of tire track patterns has an important role in both the investigation of crime scenes and the identification of vehicles involved in traffic accidents. Due to the rich texture information they have, texture features are generally used to recognize track images taken from tires. However, recognition of tire tracks taken from crime scenes has not been studied sufficiently. In this study, SIFT-based features and template matching methods were used to recognize tire track/tire track fragment images. In the experiments, fragments taken from clean tracks, dirty tracks and fragments taken from dirty tracks were matched with clean track images, and higher recognition performance was achieved compared to state of art methods.