It is difficult to accurately and rapidly partition measurement sets of multiple extended targets in cluttered environment. Hence the affinity propagation method is introduced and a novel measurement partition algorithm is proposed. First, the measurement set is preprocessed by using density analysis to remove clutters from the measurements. Second, the number and location of the extended targets is determined via competition among the measurements. Finally, state estimates are obtained by using the probability hypothesis density filter. Simulations show that the proposed algorithm offers good performance in measurement partitioning of extended target tracking with clutter disturbance. Compared with the distance partition and K-means++ methods, the proposed method effectively minimizes the computation time and retrieves the number of targets iteratively.