The invention provides an Android malicious software detection method based on a deep belief network, and the method comprises the steps of firstly, extracting the authority of an Android application software and the characteristics of a sensitive API; secondly, constructing a deep learning model by using a deep belief network (DBN), and processing the extracted features by using a deep learning model to obtain the samples representing the high-level abstract features; and then using a classification algorithm to classify the high-level abstract features outputted by the deep learning model, and distinguishing the malicious software and the normal software. By means of the deep learning model based on the deep belief network, the high-level abstract features of the Android malicious software can be better represented, and the detection effect of the deep learning model is obviously superior to that of a traditional neural network model and a traditional machine learning model.