The investigation of transparent optical layers is a growing field of application of white-light interferometry. Robust algorithms exist that extract the signal components from different layers inside a transparent structure. The separated signal contributions are then evaluated individually. Two contradicting situations have to be accounted for when low-coherence interferometry is used to measure layer structures. First, with a low NA system and a short coherence light source, the optical path difference between the layers is measured. Second, if a high NA interferometer and a long coherence light source is used, the limited depth of focus limits the correlogram width. In this case, the layer thickness is underestimated. In this paper a 2.2 μm thick reference layer is studied. This layer was measured with different interferometric systems: Michelson and Mirau interferometers with magnifications from 5x to 100x. Furthermore, light sources with different temporal coherence length were used. If lateral resolution is unimportant, the combination of a low NA measuring system and a low coherence length light source provides a larger distance between the signal contributions from different boundary layers and therefore better separation, bias correction, and higher accuracy, compared to a high NA system. The interferometer system can be calibrated by measuring the layer thickness of a small structure with respect to a substrate. Such a calibration permits performing measurements with a high NA interferometer and a low coherence light source. The main contribution of this paper is to compare and discuss results of these different options of layer thickness measurement with respect to measurement accuracy and uncertainty influences.