SUMMARYThe examination of soil fracture surfaces that are created under tensile stress may reveal a great deal about the internal structural condition of the soil. A simple technique for quantifying the roughness of soil fracture surfaces from a measurement of their topography in cross‐section is described. The technique involves calculating the standard deviation of the differences between the measured elevations of soil fracture surfaces and their corresponding running‐mean values. The standard deviation, σR, is used as a measure of the fracture surface roughness. Advantages of this technique over others are discussed.Two methods for measuring the topography of soil fracture surfaces are presented: a bisection (single transect) method, and a laser scanning (multiple transects) method. The laser scanning method is to be preferred because it requires no sample preparation and enables greater and more rapid replication. Also, fracture surfaces created by applying direct tension in the hands produced values of σRthat were statistically indistinguishable from those created using indirect tension in a loading frame. This result makes the technique of fracture surface analysis usable for both laboratory as well as field investigations of soil structure.An example of the technique is presented to illustrate the role that air‐filled pores play in the brittle fracture of unsaturated soil (air‐filled pores are closer together in drier soils and further apart in wetter soils.) A strong positive linear correlation was found between the gravimetric water content,wand σRof natural soil clods, which supports the contention that brittle fracture of unsaturated soils under tensile stress occurs at least partly because of the propagation of air‐