<p>Electric vehicle charging has a growing impact on power system carbon emissions, particularly when charging demand coincides with periods of high grid carbon intensity. While carbon-aware charging strategies have been proposed in the literature, their practical emission reduction potential under real charging behavior remains insufficiently assessed. This study provides a quantitative assessment of carbon emission reduction achieved through carbon-aware electric vehicle charging in distribution power networks. Real-world residential charging data are reconstructed at an hourly resolution and combined with time-varying grid carbon intensity to enable accurate carbon accounting. A carbon-aware smart charging strategy based on reinforcement learning is evaluated and compared with uncontrolled charging and a carbon-unaware intelligent charging strategy under identical energy delivery and network constraints. The results show that intelligent charging without carbon awareness does not reduce total carbon emissions. In contrast, carbon-aware charging achieves a statistically significant reduction in emissions of approximately 1.62% while maintaining identical delivered energy. The reduction is primarily driven by shifting charging demand away from periods with high carbon intensity, which are observed to be substantially more carbon-intensive than off-peak hours. These findings demonstrate that meaningful emission reductions can be achieved through operational coordination of charging demand without infrastructure modification. The proposed assessment provides practical evidence supporting the integration of carbon intensity signals into smart charging strategies for energy system decarbonization. The present analysis focuses on emission reduction under fixed energy delivery constraints, and electricity cost optimization is not considered in this study.</p> Graphical Abstract <p></p>