Abstract: The high-density event stream contains mass events which arrive in real time. Existing complex event detection techniques are inefficient and more memory is needed when processing high-density event streams. In order to solve these problems, a new complex event detection algorithm based on Hash structure was proposed. The effective Hash mapping structure is designed in the algorithm to store intermediate data, which improves detection efficiency. Besides, updating mechanism to delete efficiently the overdue events from the memory was also proposed. Furthermore, auto deal mechanism was improved so as to detect complex events over out of order event streams. The feasibility, efficiency and correctness of our algorithms are demonstrated by the theoretical analysis and experiments.