S. Monsell, K. E. Patterson, A. Graham, C. H. Hughes, and R. Milroy (1992) reported that high-frequency irregular words are named faster when presented in a "pure" block than when mixed with nonwords. They attributed this effect to a de-emphasis of an assembly route in the pure block. The current authors replicated this effect in Experiment 1. In Experiments 2 and 3, similar effects resulted from mixing high- and low-frequency regular words with nonwords and from mixing high- and low-frequency irregular words together. Further, in all cases, the more slowly named stimuli were named faster in mixed blocks than in pure blocks. An alternative to the de-emphasis account, which is based on strategic control of initiation of articulation, was supported in Experiment 4 by confirming the alternative account's novel prediction of a regularity effect for high-frequency words in pure blocks. Implications for the single- versus dual-route debate and for interpretations of strategy effects in general are discussed.