In English, the relationship between the written and spoken forms of words is relatively opaque, leading to proposals that skilled reading requires two procedures: (a) a sublexical grapheme/phoneme conversion process allowing the correct reading of regular words (CAT) and new or pseudowords (ZAT); (b) a lexical process necessary to read irregular words accurately (TWO) and assumed to be the dominant process for familiar words. However, it has been argued that the sublexical process may be sufficient in highly transparent languages such as Welsh. If this is the case, damage to the sublexical process may lead to more severe deficits in transparent languages due to the lack of an alternative lexical process. To test this hypothesis, we compared Welsh and English oral reading and written-word recognition and comprehension in seven bilingual stroke participants with comparably impaired pseudoword reading in English and Welsh. Performance was remarkably similar across languages. Irrespective of the language tested, words were read more accurately than pseudowords. Lexical decision and word comprehension were as accurate in Welsh and in English, and when imageability effects were present they were of a similar size in both languages. This study does not support the hypothesis that orthographic transparency determines the nature of cognitive reading processes, but rather suggests that readers develop a sight vocabulary through reading experience irrespective of orthographic transparency