We demonstrate a disturbance event recognition method based on region segmentation, which utilizes differential phase signals of a phase-sensitive optical time-domain reflectometer (ϕ-OTDR) to recognize disturbance events efficiently. The long-haul sensing fiber is divided into subsensing regions; whereas the phase signals at the two end points of the subsensing regions are subtracted, unwrapped, and differenced to represent the disturbance information. Feature extraction and classification are performed separately on the subsensing regions datasets. The experimental results indicate that the average recognition accuracy of the region-segmentation-based event recognition method is up to 92.9%. Compared to the method without region segmentation, this proposed method improves the average recognition accuracy by 8%; whereas the recognition time of three disturbance events on a 14.8-km sensing system is only 0.39 s. The proposed method provides significant support for the development of disturbance event recognition of the ϕ-OTDR sensor system.