This article investigates the parameter estimation of uncertain free-floating space manipulator. A novel integral-concurrent-learning-based adaptive strategy is proposed to cope with the above parameter uncertainties, where a set of memory data is used to construct the parameter updated law, so that the requirement on the persistent excitation condition can be removed. Besides, the proposed parameter estimation strategy does not need the information of the joint acceleration and the end-effector velocity. The Lyapunov technique is used to rigorously prove the control performance of the designed method.