Rapid automatized naming has been demonstrated as an important correlate of various reading outcomes. However, the cognitive mechanism underlying the RAN-reading relationship is not well understood. The primary goal of this study is to evaluate three major theoretical accounts for the RAN--reading relationship: phonological processing account, orthographic processing account, and speed of processing explanation. Each theoretical account would lead to different predictions on cross-language transfer of RAN to reading. One-hundred twenty nine Chinese-English bilinguals were followed from Age 4 to Age 5. They were assessed at two time points for their word reading and RAN in Chinese and English. Both concurrent and longitudinal cross-language transfers of RAN to reading were examined. The cross-language transfers from English RAN to Chinese reading were found both concurrently and longitudinally but no transfer from Chinese RAN to English reading. Our results supported the orthographic processing account. Theoretical implications are discussed.