Participants' reaction time (RT) data in a prime-probe flanker task (e.g., ABA-CAC) were analyzed in terms of the characteristics of RT distribution to examine possible mechanisms that produce negative priming. When the prime and probe were presented in the same context and the proportion of repetition-target trials (TRP) was 0.33, negative priming increased as a function of RT bins, supporting the episodic retrieval account (Neill, 1997). Manipulations that discourage the use of episodic retrieval (i.e., switching context between prime and probe or 0.00 TRP) eliminated the increase in negative priming across RT bins, even though the overall effect remained significant. These data support a dual-mechanism account in which, depending on task demands, both selective inhibition and episodic retrieval can be involved in producing negative priming. (Contains 3 footnotes, 2 tables, and 3 figures.)