Decades of repeatedly found additive joint effects of stimulus quality and frequency comprise a challenge to the IAM which predicts an interactive joint effect, but evidence for the separation of orthographic and semantic processing. We propose that the separation result from the phonology processing that could function as a “barrier” in alphabetic languages, which does not occur in Chinese language processing. This study investigated the joint effect of stimulus quality and frequency on N400 in Chinese characters which is an better index of semantic processing to test the IAM and the discrete processing of the multistage activation models.Event-related brain potentials (ERPs) were recorded during a delayed character-matching task without semantic priming in 19 healthy native Mandarin-speaking Chinese volunteers. The experiment took a 2 stimulus quality (highly blurred vs. slightly blurred) × 2 frequency: (high vs. low) within-subjects design. During the task, blurred target characters were first presented, followed by an intact probe character that either matched or mismatched the target character in identity. The participants were asked to press a button to judge whether the target, the stimulus, was in identical to the current probe regardless of their font and blurredness. There were 320 trials in a procedure and all of them were delivered in five blocks of 64 trials each and the four experimental conditions (80 trails each) were equally assigned at random to every block. The behavioral data showed a significant interaction of stimulus quality and frequency on RT with a slower RT in the highly-blurred condition than the slightly-blurred one for low frequency, but not for high frequency. The ERP data showed a large negative-going wave that was similar to N400 and higher for highly blurred than for slightly blurred stimuli near 400 ms post-stimulus onset with a broad scalp distribution. The largest blurredness effects were located in the right centro-parietal sites in a temporal window of 400 ms to 500 ms. More importantly, analyses exhibited a significant two-way interaction of stimulus quality and frequency in the centro-parietal scalp regions. Secondly, there was a larger negative-going wave in a time window of 200 to 300 ms, which was similar to N250. Its amplitude was higher for highly-blurred stimuli than for slightly-blurred stimuli regardless of frequency. However, no interactions of blurredness and frequency were significant ，and in conclusion, the N400 blurredness effect which is larger for low frequency suggests that the stimulus quality can modulate the N400 because of cascaded processing. More importantly, an interaction of the stimulus quality and frequency on both RT and N400 suggests that strong cascaded processing in Chinese characters recognition, and provides evidence for the IAM but against the multistage activation model. Meanwhile, N250 amplitude was referred to larger for highly-blurred stimuli than for the elicited slightly-blurred ones. This N250 effect of stimulus quality was interpreted as reflecting the mismatch in pattern between internal target representation and the stimulus input.