In this study, event-related potentials (ERPs) were used to trace the temporal dynamics of phonological consistency and phonetic combinability in the reading of Chinese phonograms. The data showed a significant consistency-by-combinability interaction at N170. High phonetic combinability characters elicited greater negativity at N170 than did low phonetic combinability characters, and the combinability effect was only found in the reading of high consistency characters. The results support the phonological mapping hypothesis of the reading-related N170 effect and suggest that the earlier stages of visual word recognition are shaped by the mapping of orthography to phonology even in Chinese. Moreover, our data revealed both consistency and combinability effects at P200 and N400, accounted for by the two-stage framework for visual word recognition. That is, characters with high combinability or high consistency facilitated the earlier stages of orthographic or phonological processing which were due to increased activation at the perceptual level; consequently, less positive P200 was demonstrated. In the later stages, high combinability or high consistency characters were associated with a larger semantic neighborhood, which increased semantic competition and exaggerated the N400 effect. These data support the assumption of radical-based inputs proposed by the lexical constituent model. However, the phonetic consistency effects found at N170 and P200 cannot be reconciled with the current framework of the lexical constituent model. A possible revision will be discussed.