Application of a non-destructive energy-dispersive XRF technique, involving use of analyte/Compton net peak ratios, has greatly increased the analytical potential for the study of obsidians in archaeological assemblages. Contrary to reported conclusions for obsidians from the Coso Volcanic Field of southeastern California, however, the method lacks the precision associated with powdered-specimen, thin-film, and other more rigorous techniques. Tests have shown that changing the orientation and/or placement of an archaeological artifact in a sample slot will produce widely divergent determinations of ppm concentrations for most specimens. These results indicate that discriminating between some chemical subsources at Coso cannot be accomplished reliably.