An experiment by McCann and Besner (1987) suggested that non-words that are homophonic with a high-frequency word (e.g. “keap”) are named in the same amount of time as non-words that are homophonic with a low-frequency word (e.g. “fome”), despite such pseudohomophones being faster to name than non-pseudohomophones (e.g. “feap”, “yome”). Such a result is a challenge to any model of lexical processing that places the word-frequency effect at the stage when the word is being selected from amongst all lexical entries. McCann and Besner propose a complex alternative account where frequency effects arise at a later stage of processing, a stage that is not involved in the naming of pseudohomophones. The purpose of the present study is to look for a frequency effect in the naming of pseudohomophones, controlling for orthographic factors that McCann and Besner ignored. The first experiment, using a homophone decision task, establishes a set of items that produce a clear frequency effect. These items are then used in a naming experiment which also reveals a clear frequency effect for those subjects who make use of a lexical route when pronouncing pseudohomophones. It is concluded from this that the complex model developed by McCann and Besner is unnecessary.