Several properties of foods, most notably visual perception and color, depend on the complexity of their outer surface. The so-called glistening point method based on specular reflection of an incident light beam is a simple, non-invasive optical technique used to quantify the roughness of a surface. Assuming that a surface has a macroscopic random Gaussian roughness distribution, the method determines two parameters: the correlation length (L) and the rms amplitude (sigma). Rougher surfaces have higher L and sigma values. Results show that the glistening point method is suitable for analysing food surfaces at the microscopic scale and is also useful to quantitatively follow changes in food surface roughness as during chocolate blooming or after removal of the outer wax layer in fruits. (C) 2003 Elsevier Ltd. All rights reserved.