Basic methods of studying the effect of random phase-error in large reflector antennas are discussed, with emphasis on recent developments related to Ruze's original work. It is shown that, although Ruze's basic theory has many limitations in its modified form the theory represents the most flexible technique available to date. As the surface error measured at a given instant can be resolved into independent components having more or less constant correlation intervals, Ruze's concept of correlation interval still applies.