Provided are an apparatus and a method for encoding and decoding audio signals, in which when determining a masking threshold according to a psychoacoustic model, accurate results may be obtained for a short window-based audio signal as well as for a long window-based audio signal. The apparatus for encoding audio signals according to the present invention comprises a masking threshold determining unit configured to determine, on the basis of a frame length of a first window having a divided audio signal, a masking threshold for a second window that has a different frame length from that of the first window.