This paper deals with a new iterative online dynamic identification method of robot. The robot is closed-loop controlled with a computed torque control (CTC) law that linearizes and decouples the non-linear and coupled dynamics of the robot. Usually, the parameters of the Inverse Dynamic Model (IDM), which calculates the CTC, are identified with an off-line identification method which uses the Inverse Dynamic Identification Model (IDIM) to calculate the joint force/torque as a linear relation in the dynamic parameters. This allows using linear least-squares techniques to estimate the parameters that minimize the 2-norm error between the actual joint force/torque and the calculated joint force/torque (IDIM-LS method). This method requires well tuned band-pass filtering of the joint position to estimate the joint velocity and acceleration and cannot take into account variations of the parameters (friction, payload) in CTC. The new method overcomes these 2 drawbacks. The IDIM-LS solution is periodically calculated over a moving time window to update the IDM of the CTC, and the IDIM is calculated with the noise-free data of the trajectory generator, which avoids using the noisy derivatives of the actual joint position. This method called IDIM-ILIC, works like an Iterative Learning Identification and Control procedure. An experimental setup on a prismatic joint validates the procedure with stationary parameters and with a variation of the payload.