<p>The generation and derivation of scalar implicature have always been a focal point in semantics and pragmatics. In an event-related potentials (ERPs) study, we investigated the comprehension of scalar implicature in Chinese using Rapid Serial Visual Presentation paradigm and semantic judgment task. And 26 native Chinese speakers and 28 advanced-level learners of Chinese as a second language (CSL) were presented with two utterance types including underinformative sentences (e.g., Some tigers have tails) and patently true sentences (e.g., Some animals have tails). Each stimulus type consisted of 45 sentences. The behavioral data revealed that the selection rate of scalar implicature is better than logical meaning, and remarkably higher in patently true sentences than in underinformative sentences. We also observed that native speakers have a significantly higher selection rate for scalar implicature than advanced-level CSL learners. There was no significant difference among meaning types, utterance types and participant types. The ERP results indicated that P200 amplitude in underinformative utterances was higher compared to patently true utterances. There was a significant interaction between utterance types and participant types. The post-hoc analysis showed that advanced CSL learners had higher P200 amplitudes than native speakers in the patently true utterance condition, but there was no significant difference in P200 amplitudes between the two participant types under the non-informative utterance condition. Comparing two of the utterance types, the P200 amplitude in the underinformative utterance condition was higher than in patently true utterances for native speakers, whereas no such difference in advanced-level CSL learners. We also found that utterances with insufficient information induced a steeper N400 than utterances that were patently true utterances.</p>