There is some controversy in the research literature as towhether the development of reading skills in differentorthographies varies primarily as a function of common underlyingcognitive processes (`the central processing hypothesis'), oralternatively, as a function of orthographic transparency (`thescript dependent hypothesis'). These alternative views wereexamined by studying the reading skills of 245 children in grades1--5, learning to read concurrently in English, their firstlanguage (L1) and Hebrew, their second language (L2). Childrenwere administered a non-verbal intelligence task, parallel L1 andL2 memory tasks, and word recognition and pseudoword readingtasks in both languages. Ratings of Hebrew oral proficiency wereprovided by teachers. The central processing hypothesis waspartially supported in that regardless of orthography, memoryexplained a small proportion of the variance on L1 and L2 readingmeasures. Though L2 oral proficiency was a significant predictorof L2 reading, it explained only a small proportion of thevariance. The script dependent hypothesis was supported by thefact that (a) children could read more accurately voweled Hebrew(a `transparent' orthography) than English (a `deep'orthography), (b) the developmental profiles associated withEnglish word recognition and pseudoword decoding was much steeperthan the one depicting Hebrew word recognition and Hebrewpseudoword word decoding, and (c) decoding error categories wereorthography-specific. We conclude that the two alternatives arecomplementary: When the script is less complex young childrenappear to develop their word recognition skills with relativeease, even in the absence of sufficient linguistic proficiency.At the same time, a more accurate picture of what facilitates L1and L2 reading development is enhanced when individualdifferences in underlying cognitive skills are considered aswell.