Due to the increasing traffic volume on the European railway network, the remaining fatigue life of existing steel bridges is a major concern. Several investigations demonstrated that supplementing the assessment with monitoring data enables to achieve more reliable remaining fatigue life estimations. In this paper, an event-driven monitoring system based on a wireless sensor network that consists of two functionally different components was designed and tested. Sentinel nodes, which were mounted on the track, were used for detecting approaching trains and alerting with alarm messages the monitoring nodes. These nodes, which were mounted on the bridge, started strain sensing and data recording after receiving the alarm message and went back to a power saving mode upon completion. An embedded data processing algorithm transformed the recorded raw data into a much smaller data set representing strain cycles. A test deployment on a railway bridge demonstrated that train detection and alarming was fast and reliable. The combination of event-driven monitoring and embedded data processing allowed to extend the battery lifetime of monitoring nodes to several months.