The monitoring of existing infrastructure has the potential to improve structural-condition assessment, leading to better asset management. Monitoring data, collected through load testing, help identify unknown model-parameter values, and this model-updating task is called structural identification. Then, this information is used to accurately evaluate bridge conditions, such as the reserve capacity and the remaining fatigue life of bridges, avoiding unnecessary interventions or replacement. The benefits of structural identification depend on the selected monitoring system, including the sensor types, the number of sensors, and the device locations. As these choices are usually made using engineering judgment and qualitative metrics, the selected measurement systems may be suboptimal, leading to a low cost-benefit ratio of bridge load testing. This paper proposes a methodology to rationally design measurement systems based on their cost-benefit ratio. First, optimal sensor configurations with respect to the number of sensors are assessed using the hierarchical algorithm for sensor placement. Then, a cost-benefit analysis is performed to define the optimal number of sensors. The methodology is applied to an eight-span composite concrete-steel viaduct in Switzerland, and monitoring data are taken from a static load test. Results show that a rational methodology to design measurement systems helps reduce the number of sensors without compromising the information gain, significantly improving the cost- benefit of bridge load testing.