A widely held assumption about dyslexia is that difficulties in accessing the constituent phonemes of the speech stream are responsible for specific reading and spelling difficulties. In consistent orthographies, however, the acquisition of accurate phonological recoding and phonemic awareness was found to pose much less difficulty than in English, and even dyslexic children were found to exhibit high levels of performance in phonemic segmentation (Wimmer, 1993). Nevertheless, using a rather complex phonological awareness and manipulation task (spoonerisms: MAN–HAT → HAN–MAT), Landerl, Wimmer, and Frith (1997) found support for the original position on phonological awareness deficit, as both German and English dyslexic children showed poor performance. In the present studies, the spoonerism responses of Landerl et al. were reanalyzed such that children were given credit for partially correct responses (e.g., a response of HAN for MAN–HAT). Such partially correct responses were taken to indicate full segmentation of both stimulus words at the onset–rime level. The effect of this rescoring was that the error rate dropped from 76% to 26% for the English dyslexic children and from 63% to 15% for the German dyslexic children. Even higher performance levels, although not perfect as for the age-matched control children, were found on a nonword spelling task in both groups. A second study examined the segmentation of consonant clusters in younger German dyslexic children and found performance levels of about 90% correct when memory problems were ruled out. We argue that, at least in the context of a consistent orthography (and a phonics-based teaching approach), deficits in phoneme awareness are only evident in the early stages of reading acquisition, whereas rapid naming and phonological memory deficits are more persistent in dyslexic children.