Abstract: The rock burst is a kind of large disaster in deep underground engineering, thus, it is very important to predict the rock burst. The influence factors of rock burst are numerous and their relationship is very complicated. It can not be solved by use of simple methods, Generally, based on engineering analogy and geological analysis, the clustering methods have been widely used. For the complicated environment of rock burst, this clustering problem is a very complicated fuzzy and random optimization problem, and can not be solved by use of the traditional methods very well. A new bionics clustering optimization method, ant colony clustering algorithm which is recently proposed, is introduced into the prediction of rock burst for the first time. On such a basis, a new method for the prediction of rock burst is proposed. According to analysis of the data of rock burst samples and from the engineering analogy thinking by the ant colony clustering algorithm, the rock burst can be predicted. Two examples are used to verify the new algorithm. The engineering application has proved that this new algorithm can automatically sort the rock burst samples, that the validity is very high, and that the computing velocity is rapid, so it is a very practical method.