Urban environments around the world are being highly populated by personal mobility vehicles, such as scooters or electric bicycles, which offer a new way to move around cities. Researchers from different disciplines are devoting efforts to integrate this novel vehicular paradigm into smart-city ecosystems given its advantages in terms of traffic sustainability, efficiency, and agility. However, the quick penetration of these vehicles also brings challenges and concerns related to their coexistence with other kinds of transportation systems or pedestrians, as well as the high number of accidents in which these vehicles are involved. When an accident happens, a fast and automatic detection is crucial to take quick measures, e.g., alerting emergency services. This is the main motivation of the dataset presented in this work, which provides the data captured by different sensors onboard an electric scooter under regular and accident conditions. A variety of accident kinds such as frontal collisions, lateral falls, etc. are considered, so the dataset may be valuable for the development of automatic engines to infer different riding situations.