Although studies with priming paradigm have confirmed the semantic activation of semantic radicals during the processing of Chinese phonograms in adult readers, little research has explored the processing of the semantic radical when children are involved. Most prior research in children has focused on the roles of morphological structure and positional information in Chinese compound word recognition. But note that, phonograms are the most frequently words that children learned during the whole elementary school. Besides, to achieve the meaning of a character but not to know the morphological structure and positions is the necessary process for a child learning to read. Thus, the main purpose of the current study is to investigate processing of phonograms in the developing readers. A 3 (Grade: third vs. fifth vs. adult) × 2 (Target type: transparent vs. opaque) × 2 (Prime type: form related vs. form unrelated) three factors mixed experimental design was adopted. Four conditions were formed by crossing target type and prime type. When transparent phonograms were targets, opaque phonograms served as the form-related primes; similarly, for opaque targets, transparent phonograms served as the form related primes. As refer to the two form unrelated conditions, the unrelated words served as the primes of opaque phonograms or transparent ones. Therefore, there are three types of critical stimuli: transparent, opaque and unrelated control characters. The mean character frequency, stroke number, phonogram regularity and character familiarity are balanced across these three types of characters. Besides, information constrained by phonetic radicals (i.e. radical combinability or the character frequency when it was used as a simple character) could also be kept the same across the three conditions. Eprime1.1 software was used to make programs. Each trial began with a fixation cross in the central of screen for 500ms, which was replaced by the prime stimuli (in KaiTi font) for 200 ms. The target (in SongTi font) appeared immediately after the prime and stayed on the screen until participants responded, or for 2000 ms when no responses were made. The participants were required to judge whether the targets presented on the screen were real characters by pressing the corresponding keys as quickly and accurately as possible. Firstly, grade 5 children showed faster reaction time and more accurate responses than grade 3 children on the lexical decision task, who were then better than grade 2 students. Besides, the main effect of prime relatedness was significant, with form related condition showing the slower reaction time. Significant interaction between grade, prime type and target type were also found for both reaction time and accuracy. Further analysis revealed significant interaction between prime type and target type among 3 and 5 graders, the inhibition effect of semantic radicals was greater when the target was opaque characters. By contrast, no such effects were found when Chinese 2 graders were taken into account. In conclusion, semantic information of semantic radicals can play an important role in Chinese word recognition among developing readers. However, it would associate with a child’s developing level of reading comprehension. Grade 3 and 5 readers could use the semantic meaning of semantic radicals to affect word reading, while grade 2 students didn’t show this ability. The present results provide the evidence for Taft’s model, in which researchers argued the semantic radical was presented independently in one’s mental representation.