Fatigue cracks are critical structural concerns for steel highway bridges, and fatigue initiation and propagation activity continues undetected between physical bridge inspections. Monitoring fatigue crack activity between physical inspections can provide far greater reliability in structural performance and can be used to prevent excessive damage and repair costs. In this paper, a thin-film strain sensor, called a soft elastomeric capacitor (SEC) sensor, is evaluated for sensing and monitoring fatigue cracks in steel bridges. The SEC is a flexible and mechanically robust strain sensor, capable of monitoring strain over large structural surfaces. By deploying multiple SECs in the form of dense sensor arrays, it is possible to detect fatigue cracks over large regions of a structural member such as a bridge girder. Previous studies have verified the SEC’s capability to monitor fatigue cracks under idealized harmonic load cycles with a constant crack propagation rate. Here, an investigation is performed under more complex and realistic situations to translate the SEC technology from laboratory testing to field applications—specifically, as cracking propagates under (1) a decreasing crack propagation rate, and (2) random traffic load cycles with stochastic peak-to-peak amplitudes and periods. An experimental program was developed which included an efficient data collection strategy, new loading protocols, and crack-sensing algorithms. The experimental results showed an increasing trend of the fatigue damage feature, crack growth index (CGI), under crack initiation and propagation, despite decreasing crack propagation rates or random traffic load cycles. In addition, the results also showed that the SEC did not produce false-positive results when cracks stopped growing. The findings of this study significantly enhance the SEC’s fatigue sensing and monitoring capability under more realistic loading conditions, which is a critical step toward field applications of this technology.