A two-dimensional self-calibration algorithm was developed to extract the stage systematic measurement error from a stage position measurement error. On the basis of the stage measurement error model, the algorithm got the iterative self-calibration model and the initial value by measuring five different views of an artifact on the stage, and then it established a complete iterative 2D self-calibration model. The algorithm was used to simulate a 2D stage with an accuracy of 0.2 μm. The results show that the calibration error is 0.33 nm without random measurement noises and is the same order of magnitude with random measurement noises. The actual self-calibration experiment on a stage with the given measuring accuracies of 2.98 μm and 3.22 μm in x and y directions was performed, and obtained measuring accuracies are 2.59 μm and 3.14 μm in x and y directions, respectively. All results demonstrate that the proposed algorithm has a good robustness for the random measurement noises, and it is suitable for the calibrations for precision stages or ultra-precision stages.