The rise in mega-retailers has contributed to a growing literature on buyer power and large-buyer discounts. According to Rotemberg and Saloner (1986) and Snyder (1998), large buyers' ability to obtain price discounts depends on their relative (rather than absolute) size and the degree of competition between suppliers. I test experimentally comparative statics implications of this theory concerning the number of sellers and the sizes of the buyers in the market. The results track the comparative statics predictions to a surprising extent. Subtle changes in the distribution of buyer sizes or the number of suppliers can create or negate large-buyer discounts. The results highlight the previously unexplored role of the demand structure in determining buyer-size discounts. Furthermore, the experiments establish the presence of small-buyer premia, not anticipated by the theory.