Previous studies have indicated that prenatal exposure to dioxin-like compounds (DLC) or polychlorinated biphenyl (PCB) has a negative association with neurodevelopment in school-aged children. Event-related brain potentials (ERP) can reveal subtle and specific differences in the modulation of cognitive processes that are assumed when they are associated with lower levels of prenatal exposure to DLC or PCBs. This prospective birth cohort study was conducted to examine the association between prenatal exposure to relatively low levels of DLC, PCB or methylmercury (MeHg), and ERP. A total of 55 children who were 13 years old participated in a 3-stimulus oddball task to detect P3a and P3b waves. The task required participants to respond to a target among random stimuli at two difficulty levels. The P3a amplitude reflects an automated attention capture process, and P3b reflects a voluntary attention allocation process. We analyzed DLC congeners in blood samples from four groups, including 7 polychlorinated dibenzo-p-dioxins (PCDD), 10 polychlorinated dibenzofuranes (PCDF), 4 non-ortho PCBs, and 8 mono-ortho PCBs. PCB-153 was chosen as an indicator because of its high correlation with the sum of 58 NDL (non-dioxin-like)-PCBs. MeHg exposure level was assessed by the mercury concentration in hair samples (HHg) taken during the perinatal period. The reaction time to the target stimulus during the oddball task shortened with the increasing MeHg exposure level. Furthermore, P3b latency, which reflect response decision and correlates with reaction time, was also shortened with increasing MeHg level in the difficult condition. These results are counterintuitive because shorter reaction times or rapid decision making reflected by P3 latency are generally favorable. This might be due to nutritional factors such as fatty acids, which have beneficial effects on brain development. The P3a amplitude decreased with non- and mono-ortho PCB and HHg levels, regardless of the difficulty level, and with PCDD, PCDF, and total DLC levels, especially in the difficult condition. P3b latency shortened with HHg, and P3b amplitude decreased with mono-ortho PCBs and PCB-153 in both conditions and with PCDD, PCDF, non-ortho PCBs, and total DLC in the difficult condition. In conclusion, we found an association between prenatal exposure to DLC and a decrease in both P3a and P3b amplitude, even when DLC levels were lower than in most previous studies. Additionally, our results suggest that the automated attention capture process reflected by P3a is associated with maternal MeHg exposure and that the voluntary attention allocation process reflected by P3b is associated with PCB-153. However, these results should be interpreted with caution because of the limitations on sample size, population bias, and statistical analyses.