Our research uses laboratory experiments to examine the theoretical results of competition between suppliers in an outsourcing setup. We consider a supply chain in which a single buyer needs to outsource the manufacturing of a product among N potential suppliers. The buyer allocates demand to suppliers not on the basis of price, but rather on service. We analyze the levels of service suppliers will decide to provide when competing on three different criteria specified by the buyer. For the first, suppliers compete by providing the buyer a specific service level (fill-rate), and for the second by maintaining a specific quantity of on-hand inventory. For the third criteria, suppliers compete based on a parameter designed to optimize the supply chain in favor of the buyer. Prior research and existing theory predict that the decisions will reach stability at the Nash equilibrium for all three types of competition. Theory also predicts these equilibrium points will be ordered, from competitions based on service level as the lowest and those based on the optimal criteria as the highest. Our experimental results show that the equilibrium points reached by subjects are in fact ordered as theory predicts. However, there are large and statistically significant differences between those equilibrium points and the theoretical predictions. Using the Quantal Response Equilibrium (QRE) we show that random errors can explain some of these discrepancies. Our analyses also suggest that, under optimal criteria for competition, other behavioral factors such as rival chasing and loss aversion can play an influential role.