For decoding a scalable video stream, in which the images may be decoded by groups of several images, each group being constituted by at least three levels of image temporal hierarchy, the images of the lowest level having the lowest time frequency and the images of each following level having a time frequency at least twice that of the images of the preceding level, a temporal hierarchy level being formed by at least one predetermined time interval, a time interval separating two images in that temporal hierarchy level: during the decoding of at least one temporal hierarchy level, at least one item of information is obtained representing a parameter relative to the content of the video stream over at least one time interval, and an order in the decoding of the images of each group of images is defined as a function of said information obtained.