Two methods are described for absolute measurements of the deviations from flatness of optical surfaces using a Fizeau interferometer. The first method involves averaging measurements with a reference surface that is rotated in steps. Two sets of measurements are made, one with the test surface in the optical path and the other with the test surface removed. The errors of the test surface are then obtained directly from the difference of the two data sets. The second method, which is preferable where the highest accuracy is required, involves averaging the results of intercomparisons of three surfaces, taken two at a time, with the surfaces rotated in a series of steps. The errors of each of the surfaces are then obtained from the differences of these averages. Only simple calculations are required, and random errors are reduced significantly by the process of averaging.