This paper examines exclusionary practices allegedly used by Intel to maintain its dominant position in sales of microprocessors for IBM-compatible personal computers (PCs), focusing primarily on the preferential allocation of scarce product among customers. The paper considers the period 1991-93, which encompasses the transition from the 386 to the 486 microprocessor generation and the entry of several competitors. We test for two conditions that are necessary for exclusionary practices to be successful. First, using the market definition approach that is common to antitrust analyses in the US, we demonstrate that, at the upstream microprocessor level, Intel had monopoly power in a distinct market for high-end microprocessors. In particular, pricing data show that the effects of entry were localized at the low end of Intel's product line. Secondly, we identify four conditions that are sufficient for exclusionary practices to be successful, and show that the downstream PC manufacturing stage has these characteristics. Finally, we consider potential explanations for preferential allocation that are based on efficiencies, and outline a methodology for calculating the effect of exclusionary practices on the prices received by two of Intel's competitors