Abstract: Aiming at the characteristics of dumping collapse in Guizhou Province, factors (slope, rock structure, the cavity state, dip of the master structural plane, unloading and relaxation, the cutthrough extent of cracks at the trailing edge, degree of softness and hardness of the base) which control, influence the occurence of disaster and are easy to collect at field were selected as indexes. In order to lay the foundation for easy field identification of the potential collapse, based on RES rock engineering systems theories, through using the BP artificial neural network model to the intereactive relationship matrix of the indexes, the paper studied the influence of the intereaction of indexes on the stability of the potential collapse, determined the weight of evaluation indexes, and the importance of the evaluation indexes in the overall stability was sorted. The study found that rock structure, the cutthrough extent of cracks at the trailing edge have strong influence on the potential dumping collapse.