A gravel-bed stream's grain size distribution plays an important role in determining a river's depth, sediment transport rates, and stream bed stability as well as the survival rates of mussels and salmonids. Unfortunately, the material found in gravel-bed rivers exhibits vertical stratification as well as spatial variation in the horizontal direction and is difficult to sample. Previous research has largely dealt with the ability of grid, areal, and bulk sampling techniques to sample a single spot within a river. Little has been done in characterizing an entire river reach. Of the methods suggested, none is adequate because they are either inherently biased or are incapable of describing the spatial variations within a sampled region. The present research proposes a method that overcomes these problems. It shows that a single large grid sample, or composite grid sample, can be used to obtain an unbiased estimate of an area's overall grain size distribution at a known accuracy level. It then suggests that the arithmetic mean is a suitable parameter to characterize the coarseness of individual sediment deposits within a sampled area. Thus, by recording the size and location of each stone taken in the composite grid sample one can use statistical hypothesis testing to systematically analyze local means throughout the sampled area and locate sediment boundaries. Once the boundaries are located, stones from the composite grid sample falling within the boundaries of a particular deposit can be analyzed as separate grid samples representative of the individual deposits present and describe the local variability. / Master of Science