Previous studies have shown that physician’s pain empathy was significant reduced compared to that of non-physicians, which could be manifested by their suppressed early emotional sharing processes indexed by N1 and late cognitive evaluation processes indexed by the P3 or LPP components. However, it remains unclear whether medical undergraduates at preservice education stage would have the same problem as well, and whether their pain empathy would be attenuated by repeated exposures. The investigation of these two issues would not only help us deeply understand the adaptability and plasticity of social empathies, but also provide useful empirical and theoretical references for the health protection and professional development of physicians. This study used the event-related potentials (ERPs) method to compare the cognitive processes of pain empathy between medical and non-medical undergraduates. Furthermore, we also compared the possible influences of repeated exposures on pain empathy between these two groups. During the study, a total of 25 medical undergraduates and 25 non-medical undergraduates (3 non-medical undergraduates was excluded because bad EEG recording) were tested by a three-stage experimental paradigm: pre-exposure test, repeated exposure of pain in the middle stage, post-exposure test. In both pre-exposure and post-exposure test stage, the classical picture observation paradigm (Decety et al. 2010) was used. While in the repeated exposure test stage, a free-view habituation task (Mastria et al. 2017) was used. That resulted a 2 (participants: medical student vs control group) × 2 (stimuli: pain pictures vs non-pain pictures) × 2 (task stages: pre-test vs post-test) experiment design. ERP components associated with pain empathy, such as N1, N2, P3, and LPP components, were measured and analyzed accordingly. Our behavioral results showed that both the medical and non-medical undergraduates exhibited similar pain empathy effect, and neither group showed significant difference between pre- and post-repeated exposure. However, ERP results showed that the temporal processing characteristics of the pain empathy processes in medical undergraduates were significantly different from that of non-medical undergraduates. Specifically, compared with the control group, medical undergraduates failed to exhibit similar pain empathy effects indexed by significant N1 effect between pain and non-pain stimuli, but showed the comparable increased reinforcement on late cognitive evaluation processing indexed by enhanced LPP amplitude to pain stimuli. Regarding to the repeated exposure effect, pain empathy for medical undergraduates were resistant to the repeated exposure by showing comparable enhanced LPP components to pain stimuli in both pre- and post-exposure test stages. On the contrary, pain empathy for non-medical undergraduates were significantly reduced after the repeated exposure by showing reduced LPP components to pain stimuli between pre- and post-exposure test stages. To sum up, our study showed that, although medical preservice education suppressed the early emotional sharing processes of pain empathy, medical undergraduates could maintain their pain empathy level through investing more controls and mentalizing resources at late cognitive evaluation stage. Additionally, compared to the non-medical undergraduates, their pain empathies are less influenced by repeated exposures. Taken together, these findings are not only of great significance for understanding the occupational psychological characteristics of medical staffs, but also demonstrate that pain empathy shaped by knowledge learning can resist repeated exposure or habituation.