Although much has been learned about the nature and etiology of impairments in sublexical translation skills, less is known about the difficulties of children who are poor at recognising words as whole orthographic units. In the research described here, an attempt was made to look closely at the process these poor readers go through when trying to acquire new lexical information. Groups of poor readers were required to learn the pronunciations of a set of “irregular” nonsense words over a series of training sessions. There was an imperfect correspondence between the orthographic and phonological forms of these items, so grapheme‐phoneme conversion rules could not be used (e. g.,machtwas pronounced “mot”). The results indicated that this learning task distinguishes between poor readers with a specific lexical deficit and those with a different type of reading difficulty. Ways in which this kind of task might usefully be employed in future research in this area are