A multichannel dynamic compression algorithm for use in hearing aids is introduced which uses a loudness estimation model to adapt the frequency-gain characteristic to signals of arbitrary spectral shape and level. The model was incorporated into the algorithm described in [2] and has been compared with an optimally fitted linear system (frequency shaping only) in terms of subjective transmission quality and speech intelligibility in quiet and in noise. The results show an advantage of the proposed algorithm in comparison to linear frequency shaping, especially when the broad dynamic range of signals in everyday life is considered.