ABSTRACT: The pebble count procedure (Wolman, 1954) is the measurement of 100 randomly selected stones from a homogeneous population on a river bed or bar, which yields reproducible size distribution curves for surficial deposits of gravel and cobbles. The pebble count is widely used in geomorphologr (and increasingly in river engineering) to characterize surficial grain size distributions in lieu of bulk samples, for which adequate sample sizes become enormous for gravels. Variants on the original method have been proposed, one of which, the so-called ‘zig-zag’ method (Bevenger and King, 1995), involves sampling along a diagonal line and drawing data points from many different geomorphic units. The method is not reproducible, probably because it incorporates stones from many different populations, and because an inadequate number of grains is sampled from any given population. Sampling of coarse bed material should be geomorphically stratified based on the natural sorting of grain sizes into distinct channel features. If a composite grain size is desired, the areas of the bed occupied by different populations can be mapped, pebble counts conducted on each, and a weighted average distribution computed.