Based on a simple model of lithic procurement, reduction, and use, we generate predictions for patterns in source diversity and average distance-to-source measurements for flaked stone assemblages left behind by small-scale and residentially mobile populations. We apply this model to geochemical data from obsidian artifacts from three regions in western North America. As predicted, results show markedly different patterns in the geochemical composition of small flakes, large flakes, and formal tools. While small flakes and tools tend to have greater source diversity and are on average farther from their original source, the large flake assemblage is composed of fewer and closer sources. These results suggest that a failure to include very late stage reduction (e.g., pressure flakes) and microdebitage in characterization studies may bias interpretations about the extent of residential mobility and/or trade patterns because more distant sources will be underrepresented.