Effects of 4-dimension variational data assimilation (4DVAR) with multifold observed data on the typhoon track forecast are studied, by using the MM5V3 model, the RTTOVS-5 model and their adjoint models. The data used for assimilation include large-scale background fields data,bogus data, cloud-derived wind data, satellite inverse data,and high resolution infrared radiation sounder data (HIRS).There are 5 typhoon cases are used to perform the numerical experiments and assimilation experiments. The numerical results show that with 4DVAR the initial fields can be greatly improved and the initial typhoon structure can be clearly described.