Abstract: This paper studies the method of direct position determination (DPD) using merely Doppler frequency shifts (DFS) in a distributed Multiple-Input Multiple-Output (MIMO) radar system. In general, the conventional two-step localization methods have to firstly extract some positioning parameters, such as time information and frequency information, from the received signals and then estimate the final result by solving the relation equation between these parameters and the target position. However, these procedures are suboptimal due to the first step disregards the constraint that the extraction of all measurement parameters must be relative to the same target position. To overcome this problem, this article presents a direct position determination algorithm based on maximum likelihood (ML) criterion and this algorithm can process all received data centrally without any extraction procedure, and it is able to estimate the target position by one-step. The simulation results show that the proposed algorithm can achieve a better estimation than the two-step localization method in a low signal to noise ratio (SNR) condition, and attain the identical estimation accuracy with the two-step method in a high SNR condition.