Three experiments were conducted to compare the development of orthographic representations in children learning to read English, French, or Spanish. Nonsense words that shared both orthography and phonology at the level of the rhyme with real words (cake-dake, comic-bomic), phonology only (cake-daik, comic-bommick), or neither (faish, ricop) were created for each orthography. Experiment I compared English and French children's reading of nonsense words that shared rhyme orthography with real words (dake) with those that did not (daik). Significant facilitation was found for shared rhymes in English, with reduced effects in French. Experiment 2 compared English and French children's reading of nonsense words that shared rhyme phonology with real words (daik) with those that did not (faish). Significant facilitation for shared rhyme phonology was found in both languages. Experiment 3 compared English, French, and Spanish children's reading of nonsense words (dake vs. faish) and found a significant facilitatory effect of orthographic and phonological familiarity for each language. The size of the familiarity effect, however, was much greater in the less transparent orthographies (English and French). These results are interpreted in terms of the level of phonology that is represented in the orthographic recognition units being developed by children who are learning to read more and less transparent orthographies.