Abstract: The traditional methods for spacecraft autonomous navigation need several sensors, such as star sensor, infrared horizon sensor and magnetometer, to collect navigation data. As a result the load of spacecraft will gain in weight, size and power. Based on the advantages of multi-field of view (FOV) star tracker, an autonomous navigation method was proposed which used multi-FOV star tracker (MFST) to image the star and the earth respectively and got the orientation vectors of them. Combining with the orbit parameters of the spacecraft and the layout of the MFST, a mathematic model of the earth imaging was set up to implement the earth edge images in every single FOV. The Steger method was used to determine the earth edge in the images. Considering the earth oblateness, the orientation vector of the earth will be obtained through circle-fitting the earth edge points in each FOV. With the configuration that one FOV measures the navigation star and the other two FOV measures the earth edge, the autonomous orientation based on the starlight angle is simulated and the result indicates that the accuracy of the earth vector and the spacecraft position respectively reaches 0.017 2(1) and 190 m(1)。