A method for decoding a multilayer video signal, according to the present invention, is characterized by: inducing the number of active references of a current picture, in consideration of whether the maximum number of active reference pictures of the current picture in a current layer is limited; obtaining a reference layer identifier based on the number of active references, determining an active reference picture of the current picture using a reference layer identifier, generating a reference picture list including the active reference pictures, and performing interlayer prediction of the current picture using the reference picture list.