The manufacturing process of CNC machine tool is a typically nonlinear, timevarying and uncertain dynamics vary process because its position servo system is influenced by varying factors, such as machining environment, component shape and mechatronic properties. Therefore, it is challenging to build the precise dynamics model. Based on the equivalent data model of the controlled object, a datadriven adaptive iterative learning control approach was proposed by utilizing the repetitive operating pattern of a batch of same components. Through the dynamic linearization method on the iteration axis, the automatic updating of the learning control gain was achieved by minimizing the control objective function with only requirement of the input and output data of the CNC system. The proposed approach overcome the problem of fixed gain for the traditional P-type iterative learning control, its convergence was guaranteed through a rigorous theoretical analysis.The simulation results show that the mean absolute error and maximum absolute error of the datadriven adaptive iterative learning control method reduce by 46% and 56% than those of the traditional P-type iterative learning control method.