This paper presents an analysis of the effects of microroughness of the ocean bottom on a sonar signal. The results best apply to features where the roughness amplitude is less than one‐quarter of an acoustic wavelength such as with ripples, beds of rocks, and nodules. The shape of the probability density function (PDF) of the echo envelope is examined in terms of the rms roughness and a new parameter, the correlation area of the bottom. The area is equal to the product of thexandycorrelation distances along the floor. The PDF is shown to be extremely sensitive to small changes in the roughness. Furthermore by determining the rms roughness from standard coherent reflection measurements, the correlation area may be extracted directly from the PDF. Thus both vertical as well as lateral information is obtainable from sonar data. The technique can be used to discriminate between different types of bottoms that may have the same roughness but different correlation areas, for example a floor with ripples versus one with rocks or nodules. The analysis combines (1) a general statistical model employing the Rice PDF [S. O. Rice, inSelectedPapersonNoiseandStochasticProcesses, edited by N. Wax (Dover, New York, 1954), pp. 133–294], and (2) a theory originated by Eckart [C. Eckart, J. Acoust. Soc. Am.25, 566–570 (1953)]. The analysis is applied to sonar data collected from the continental shelf near Cape Hatteras, North Carolina. The results are consistent with the known characteristics of the area.