A wide range of studies have shown that executing the other secondary task during retrieval phase has influence on implicit memory (repetition priming). But, we still don’t know the mechanism of the interference. There are few relative studies discussed and verify the mechanism of interference effects on implicit memory during retrieval. Therefore, this study was designed to investigate the mechanism of interference with event-related brain potential (ERP) technique. The effect of interference during retrieval was assessed by comparing two divide-attention (DA) conditions, on which participants carried out an interference task (even-odd decision) and a memory task (lexical decision) successively, with a full-attention (FA) condition, on which participants performed only the memory task or interference task. We combined lexical decision task with go/no-go task for investigate mechanism of interference. In a trial, there was a digit presented at the center of screen, followed by a lexical. Participants had to judge whether the digit was odd or even and the lexical was word or pseudo-word. The upcoming task was indicated by a task cue. A digit with underline or a digit without underline indicating whether participants were required to respond to the digit (go vs. no-go). Participants were instructed to make response to corresponding items by pressing keyboard. They were asked to perform both tasks as quickly and accurately as possible. The Reaction Time (RT) and Accuracy data in retrieval phase were recorded, in order to assess priming effects, the effect of interference and interference task costs. The ERP data also were recorded and analyzed. The results showed that, priming effects were obtained in the FA condition and no-go condition but disappeared after go trials. There was non-significant difference in Accuracy. The ERP data are the same as behavior data. The N400 old-new effect was observed in FA condition and no-go condition but disappeared in the go condition. In addition, we found that the situation of P200 was the same as N400.The old word produced bigger amplitude P200 than new old in FA condition and no-go condition. But it was not observed in go condition. In conclusion, results from the current study revealed that implicit memory priming could not be regarded as an automatic form of retrieval with ease. Inertial activation of the pre task will impact implicit memory during retrieval. But when inertial activation was inhibited, the implicit memory would be recovery. So, the reason that the implicit memory is interfered during retrieval is interference task and implicit memory task competes for limited cognitive resources. It provides evidence for the view of “general cognitive resources compete”. On the other hand, the ERP data also provide evidence for implicit memory will be interfered when cognitive resources are insufficient during retrieval. In addition, we found that P200 might indicate a kind of perception priming. It also was interfered when cognitive resources are insufficient during retrieval. We believe that implicit memory may not be completely automatic processing. And, retrieval phase need more cognitive resources compared with encoding processing.