There is a long-standing argument about whether words or character morphemes are the functional units in reading Chinese. We propose a Character-Word Dual Function (CWDF) model of reading Chinese in which both characters and words are functional units that contribute differentially to orthographic and meaning processes in reading Chinese. Two masked constituent priming experiments examined one of the model's predictions by testing orthographic and meaning priming effects in reading Chinese compounds. The meaning transparency of the prime (a constituent character) and the target (a compound word) was manipulated to distinguish meaning effects from orthographic effects. In transparent pairs, the prime and target were orthographically and semantically related (e.g., [characters omitted] [home-family]); in opaque pairs, the prime and target shared orthography only (e.g., [characters omitted] [home-guy]). Exposure durations of the prime were varied (50ms and 216ms) to allow orthographic and meaning priming, respectively. We manipulated word frequency to reflect the accessibility of word-level representations. The results are as follows: (1) Sharing character level orthography facilitated word reading for both high- and low-frequency words, supporting the assumption of an important role for the character level in orthographic processes. (2) The meaning priming effect was found with the longer exposure duration and only for low-frequency words. High-frequency words produced rapid access, which was not enhanced by the meaning priming of its constituent characters, suggesting that the word functions as the primary meaning provider in reading. These findings are congruent with the CWDF model, which posits that characters are the fundamental orthographic units of written Chinese and that words are primary meaning providers.