Introduction: Event-related brain potentials (ERPs) provide important information about the sensitivity of the brain to process varying risks. The aim of the present study was to determine how different risk levels are reflected in decision-related ERPs, namely the feedback-related negativity (FRN) and the P300. Materials and Methods: Twenty participants conducted a probabilistic two-choice gambling task while an electroencephalogram (EEG) was recorded. Choices were provided between a low-risk option yielding low rewards and low losses and a high-risk option yielding high rewards and high losses. While options differed in expected risks, they were equal in expected values and in feedback probabilities. Results: At the behavioral level, participants were generally risk-averse but modulated their risk-taking behavior according to reward history. An early positivity (P200) was enhanced on negative feedbacks in high-risk compared to low-risk choices. With regard to the FRN, there were significant amplitude differences between positive and negative feedbacks on high-risk choices, but not on low-risk choices. While the FRN on negative feedbacks did not vary with decision riskiness, reduced amplitudes were found for positive feedbacks in high-risk relative to low-risk choices. P300 amplitudes were larger in high-risk decisions, and in an additive way, after negative compared to positive feedback. Discussion: The present study revealed significant influences of risk and valence processing on ERPs. FRN findings suggest that the reward prediction error signal is increased after high-risk decisions. The increased P200 on negative feedback in risky decisions suggests that large negative prediction errors are already processed in the P200 time range. The later P300 amplitude is sensitive to feedback valence as well as to the risk associated with a decision. Thus, the P300 carries additional information for reward processing, mainly the enhanced motivational significance of risky decisions.