The Short-Circuit Hypothesis is one of the best-known theories in L2 (second language) reading. It proposes that the L2 reader has to reach a threshold level of L2 language proficiency in order to transfer his/her reading skills to L2 reading; otherwise insufficient knowledge of the L2 will “short-circuit” the readers’ reading system. In the 1980s, against the background of this hypothesis, the issue of the relation between L1 (first language) reading and L2 reading has been a matter of theoretical debate in many publications as “a reading problem or language problems”. Accumulative evidence was in support of the hypothesis. However, discussions came to a stop in the 1990s, and research on L2 reading procrastinated. Many key issues proposed by this hypothesis, however, remain to be resolved. A major problem with the previous research is that many studies were methodologically test-based and product-oriented. Little research was devoted to the investigation of the psycho-cognitive nature of the threshold effect and the short-circuiting of readers’ reading systems. The present article attempted to approach the short-circuit hypothesis and examine the threshold effect by viewing L2 reading as a complex psycho-cognitive process. More specifically, using the Structure Building Framework by Gernsbacher (1990), it examined how L2 readers built mental representations of the L2 texts they were reading and what cognitive processes were involved in the threshold effect for L2 reading. Two groups of participants with different levels of English proficiency took part in an online task of L2 narrative processing. During the test, they read 30 stories in both English and Chinese which were experimentally manipulated to have different kinds of story character: rementioning the original character, introducing a new character, and presenting neutral information. After reading each paragraph, they responded to a test probe, which was designed to test whether they could keep track of story characters and thus follow the causal chain in the stories. Results indicated that the L2 group with a high level of English proficiency successfully transferred their comprehension skills from L1 to L2. The data of the L2 narrative comprehension further indicated that they efficiently suppressed the noisy information (irrelevant story characters) during comprehension and built “networked, cohesive” mental structures with few substructures. For the L2 group with a low level of English proficiency, a different pattern of results was found. They failed to transfer their L1 comprehension skills to L2 reading. It was also shown that, although they could suppress the noisy information during L1 comprehension, they failed to suppress the irrelevant information during L2 narrative comprehension, which resulted in “bulkier, less cohesive” mental structures of the L2 texts. Based on these results, this article proposes that the transfer of L2 reading comprehension skills is associated with the readers’ mechanism of suppression. It was the inefficiency of their mechanism of suppression that hampered low-level readers’ transfer of L1 comprehension skills to the L2.