Adaptive neuro-fuzzy inference system combines the interpretability of fuzzy reasoning and the ability of adaptability and self-learning of neural network, overcoming the enormous difficulties for slope stability analysis caused by the uncertainty of rock slope. At the same time, for the reasonable of parameter setting in fuzzy inference system, evaluation model of adaptive neuro-fuzzy inference based on genetic algorithm is established. It was shown that the results of GA-ANFIS model are consistent with field condition, which makes it be an effective method for slope stability evaluation.