Inhibition in task switching is inferred from n–2 task repetition costs: Slower response times and poorer accuracy for ABA task switching sequences compared to CBA sequences, thought to reflect the persisting inhibition of task A across an ABA sequence. Much work has examined the locus of this inhibition effect, with evidence that inhibition targets response-selection processes. Consistent with this, fits of the diffusion model to n–2 task repetition cost data have shown that the cost is reflected by lower estimates of drift rate, suggesting that inhibition impairs information processing efficiency during response selection. However, we have shown that the n–2 task repetition cost is confounded with episodic retrieval effects which masquerade as inhibitory costs. The purpose of the current study was to conduct a comprehensive analysis of diffusion model fits to new data within a paradigm that controls for episodic interference. Across 4 experiments (total N = 191) we find evidence that the reduction of drift rate for n–2 task repetition costs is only evident under conditions of episodic interference, and the cost is absent when this interference is controlled for. In addition, we also find evidence that episodic retrieval influences task preparation processes and response caution. These findings provide important constraints for theories of task switching that suggest inhibition selectively targets response selection processes.