In a method and apparatus for determining the height of a plurality of spatial positions on a surface of a specimen, a light beam is projected on the surface. The surface is scanned along an optical axis in different scanning positions. The light reflected by the surface is detected in scanning positions with a spatial pattern having corresponding spatial pattern positions. From the detected light for each spatial position of the surface, an envelope curve of intensity values corresponding to scanning positions is determined. A maximum of the envelope curve and its corresponding scanning position being representative of the height of the spatial position of the surface is selected. The spatial pattern is moved in a sequence of 2n steps (n>2) in a first and a second spatial direction over a distance of ¼ and 1/n pattern wavelength, respectively.