The neural network algorithm in this paper is applied to the ocean surface wind speed retrievals. Firstly, the Ku band backscattering coefficient (σ 0 Ku) is considered as the input parameter to retrieve the wind speed and the retrieval precision reaches 1 m/s (root mean square error) for HY2 altimeter. Secondly, by introducing the Ku-band significant wave height (swh ku), C-band backscattering coefficients (σ 0 C) and C-band swh (swh C) as input parameters to inverse wind speed, the retrieved results show that the multi-parameter algorithm introduced in the neural network can effectively improve the wind speed retrieval accuracy. The wind speed is not only relative to σ 0 Ku , but also to σ 0 C , swh ku and swh C. The neural networks algorithm is available for HY2 altimeter wind speed retrieval.