Grounded in a follow-up study among children who lost one or both parents to HIV in central China in the early 2000s, we conducted an event-related potentials (ERPs) experiment to explore the effect of social exclusion on trust and the corresponding neurophysiological mechanism among youth orphaned by HIV/AIDS (“AIDS orphans”). A sample of 31 AIDS orphans (26.16 ± 3.34 years old; 15 female) and 32 age and development status matched controls (25.02 ± 3.45 years old; 14 female) participated in the study. They were all assigned to play Cyberball, a virtual ball-tossing game that reliably induced social exclusion (15 orphans, 16 controls) and inclusion (16 orphans, 16 controls). Then, they played the Trust Game by taking the role of trustor with their electroencephalograms (EEGs) being recorded during the game. In the Trust Game, each participant was required to decide whether to trust their partners in over 150 trials (decision-making stage). The partner’s reciprocation strategies were pre-programmed by the experimenter (with an overall reciprocating rate of 50%). All participants were provided with post-decision feedback about the outcome of their decisions (gain or loss of game points) in each trial (outcome evaluation stage). We analyzed their behavioral responses at the decision-making stage and ERP components at the outcome evaluation stage. Behavioral results showed that the proportion of orphans choosing trust was significantly higher than the controls, and the trust ratio of the orphan exclusion (OE) group was significantly higher than that of the orphan inclusion (OI) group, control exclusion (CE) group, and control inclusion (CI) group. Furthermore, the response time of the OE group was significantly shorter than that of other groups. ERP results indicated that the amplitude of the feedback-related negativity (FRN) in the OI group was significantly more negative than that in the CI group with loss feedback, while there was no significant difference between the OE and OI groups. Similarly, the P300 amplitudes following outcome feedback were larger in the CI group than that in the OI group with gain feedback and had no significant difference between OE and OI.