Abstract: In order to achieve adaptivity, parallelism and high speed, neural network has been applied and developed rapidly in the field of data fusion recently. Many new algorithms based on neural network have been proposed. Some of them proposed after the 1980 s and their applications in the data level, feature level and decision level fusions are described in some detail. Some modified methods and emulation results to these algorithms are also proposed. The development of neural network in data pretreatment for data fusion is introduced. Finally, prospects of neural network are described.