Driver emotion significantly impacts traffic safety, driving behavior, and driving experience, often being directly influenced by different driving scenarios. To explore this relationship, we present EmoRoad, a multimodal dataset capturing emotional, physiological, and behavioral responses under diverse driving scenarios. These scenarios are defined by three dimensions, each with two conditions: road scenario (urban/suburban), traffic density (jam/flow), and weather (sunny/rainy). The combination of these factors results in eight representative driving scenarios. Data were collected from 50 participants (30 female, 20 male; aged 18-67), including first-person driving videos, facial videos, EEG signals, eye-tracking data, steering wheel touch data, vehicle dynamics, and emotion annotations. EmoRoad offers a rich resource for research on emotion recognition, behavior modeling, and the effects of driving context on emotion, with potential applications in intelligent transportation systems and affective computing.