The results of research over 30 years have found a reliable morpheme effect on word recognition, suggesting that morphemes are involved in lexical access. Although evidence supports a unique and independent role of morphemes in word recognition, it remains controversial whether early morphological processing relies purely on surface orthographic or morphemic meaning. Previous researchers commonly address the issue by investigating the semantic transparency which compare the strength of morphological priming produced by transparent words (e.g., snowman-snow, morpheme and whole words share both form and meaning) and opaque words (e.g., department-depart, morpheme and whole words share form only). However, regretfully, current results are confusing, with some studies reporting that the prior exposure to both transparent and opaque words could facilitate subsequent target words and the strength of facilitation in two conditions are equivalent, suggesting morphemic form was only sufficient for morphological priming regardless of the morpheme meaning. However, others assume the strength of facilitation produced by transparent words is stronger than opaque ones, which indicates morpho-semantic processing may start at the early stage of morphological processing. The majority of Chinese characters are compounds, which adopt a morph-syllabic system that each Chinese character corresponds to a morpheme, in addition, morphemic decomposition may be relatively easy and possibly automatic because morpheme boundaries are marked by space which make it more adaptive to extract morpheme meaning during Chinese word recognition. It is more apparent when Chinese compounds containing ambiguous morphemes (e.g., 月, meaning moon or month) may have different meanings which only share the same form can be a better choice to examine morpho-semantic processing, it is worth noticing that the ambiguous morpheme we use here does not distinguish dominant meaning from subordinate meaning that the relative frequency of the alternative interpretations of ambiguous morphemes could be similar. Therefore, the present study was conducted to explore morpho-semantic activation during the early stage of Chinese word recognition. In the research, masked priming paradigm is used and lexical decision task adopted, combined with ERPs technique. Two factors were manipulated: the type of target words containing ambiguous morpheme one of the two alternative meanings and the type of prime words either containing ambiguous morpheme with two meanings or unrelated to the target words. Behavioral results reveal that only target words sharing the same morpheme both form and meaning with prime words could facilitate the recognition of target. ERP results further confirmed behavioral results which showed the interaction between prime words type and target words type as reflected by the N250 and N400 effects, only when target words share the same morpheme both form and meaning with prime words would the change be elicited of N250 or N400, the effect of N250 mainly observed within frontier region, while the effect of N400 was more widely observed in all regions compared to N250 which indicated that the role of ambiguous morphemic meaning in word recognition was done in two stages reflected on N250 and N400. In result, morpho-semantic processing was involved in the early stage of morphemic processing. Moreover, it adds the new evidence to psychological meaning of N250, which was related to morpho-semantic processing.