The stability prediction of rock slopes is a key and complicated problem. We took 24 typical rock slopes in hydropower engineering regions in China as examples, to build distance discriminant analysis (DDA) model with posterior probability for slope stability prediction. The 5 combined indexes, i. e, slope rock mass quality coefficient (SRMR), orientation coefficient of structural plane (F), modified coefficient of structural plane types (λ), slope height coefficient (ℰ) and modified coefficient of construction methods (Kw), were used as the discriminant factors. The analysis results show that DDA method has a lower error rate, the average accuracy rate is 91.67%. By using the DDA method, we can obtain the posterior probability belonging to the different classification as well as the classification of slope stability. Therefore, the method may provide a new way for slope stability prediction.