A single object tracking algorithm based on spatio-temporal feature enhancement (OSTrack-ST), which was built on the one-stream tracking network OSTrack, was proposed to address the common issues of occlusion and scale variation in complex motion scenes and enhance the performance of single object tracking algorithms in utilizing temporal feature information and expressing object spatial features. For spatial feature enhancement, a multi-head spatial association attention mechanism including the spatial attention and the multi-head context association attention was proposed to enhance the model’s ability to express global and local spatial features, and effectively improve the model’s ability to capture object features in dynamic environments. For spatio-temporal feature enhancement, a spatio-temporal template update strategy based on temporal drift prediction was proposed, which used spatial position prediction results to control template updates over time and enhanced the robustness and accuracy of the model in long-term sequential tasks. Experimental results demonstrated that the proposed algorithm achieved tracking success rates of 70.5%, 73.7% and 68.7% on the LaSOT, GOT-10k and SportSOT datasets while the running speed was over 49 frame per second. The overall performance of this algorithm was better than that of other tracking algorithms such as EVPTrack.