Abstract: In order to improve the accuracy and reliability of the nuclear accident radiation field dose rate assessment in early phase, a method based on data assimilation theory was proposed. Using data assimilation theory, state vector of radiation field was defined, and forecast model operator and error covariance matrix for radiation field were established. To achieve optimal expression of the radiation field dose rate, both model predictions and measurements were taken into consideration by assimilation algorithm. In order to verify the reliability and accuracy of data assimilation methods, both numerical simulation experiment and experiment related on utilizing dose rate data measured at Fukushima were tested. The results show that under the condition of numerical simulation, the relative root mean square error of radiation field can reduce from one magnitude to about 50% caused by source term. By assimilating measurements data of Fukushima, the results are close to measurement data, and the relative root mean square error is about 20%. The above results show that the method of data assimilation can effectively reduce the uncertainty of radiation field prediction.