Building a knowledge base is one of the most important tasks in development. New content can be entered into semantic memory not only through direct experience (e.g., explicit instruction, observation), but also through self-generation resulting from a number of logical processes, including deduction, induction, and analogy. Another means of extension of semantic memory is self-generation of new knowledge through integration of information acquired in two (or more) separate yet related learning episodes. Previous studies mainly investigated knowledge extended through integration remains accessible after delay, little is known about the time course over which newly self-derived information through integration and the effects of prior knowledge on undergraduate students’ self-generation newly knowledge through integration. The present study aimed to examine the cognitive processing and the time course of self-generated through integration of separate but related facts pair by using the event-related potentials (ERPs) technique of high temporal resolution and to the status of semantic knowledge. Considering the role of prior knowledge experience, this study distinguishes knowledge integration into instant integration and delayed integration. There were five categories of sentences in experiment: (a) well-known facts; (b) derived new facts through integration of separate immediate stem facts presented during encoding; (c) derived new facts through integration of separate delayed stem facts presented during encoding; (d) derived inference facts through transitive facts presented during encoding; (e)novel control facts unrelated to any previously encountered. At the encoding task, subjects read each of 180 sentences presented on a computer, then judged when in school the facts in the sentences should be learned: primary, middle school, or college. After a break of 5 minutes, participants were tested generation of new information through integration of the previously presented stem facts during encoding. ERPs were collected as subjects read each of 100 sentences. To permit observation of anticipated change over time in the neural status of the self-derived information, each sentence was presented twice. On the behavioral level, we found that the main effect of the number of presentations is significant, and the average reaction time of all fact types on the second presentation is significantly faster than the first presentation; the main effect of fact type is also significant. The response time of well-known facts is significantly faster than that of immediate integration, delayed integration and novel facts. On the ERP level, N400 responses to integration facts differed from responses to transitive reasoning facts and novel facts, and did not differ from responses to well-known facts, suggesting that the newly self-derived newly information through integration had been incorporated into the semantic knowledge base. In addition, from the analysis results of time course, we also found that during the second presentation, the amplitude of the delayed integration facts decreased and moved closer to the well-known facts, indicating that through integration of separate delayed stem facts had been occurred later. Combined with behavioral and ERP findings, these results showed that the time course of self-generated new knowledge through integration of separate but related facts pair and the prior knowledge experience does have an impact on integration performance. The results indicate that self-derived newly information through integration can be incorporated into the semantic knowledge base and knowledge experience will affect the time course of integration. The finding helps explain the richness of semantic memory and also further our understanding of the basic cognitive processes involved in extension of knowledge.