Nowadays characterization of surface roughness using three-dimension (3D) methods and parameters become more and more important. This type of surface characterization provides a more complete view on the surface qualities, since the surface roughness is viewed as a spatial object. Within the last ten years intensive work is being carried out on the development of 3D roughness standards, therefore it is necessary to agree on a unified approach in the assessment of surface roughness 3D parameters. To make possible application of the ISO/DIS 25178 standard being developed one needs information on the determination of 3D surface roughness parameter precision, such as number of measurements, dimensions of measurement areas and their disposition on the measured surface. The present work provides the surface roughness assessment principles when measuring the 3D roughness of coated surfaces by means of contact-type measurement devices, based on the regularities of the theory of probability. The work considers one of the basic parameters of roughness – the surface arithmetic mean deviation Sa, describing surface by the help of a normal random field at a monotonously decreasing two-dimension correlation function type.