A value of information based approach for optimal structural health monitoring design using ultrasonic guided-waves is introduced in this paper. The method trades-off information gained from the ultrasonic data and costs associated to the number of sensors, while accounting for modeling and material uncertainties. The optimization of the position of sensors is carried out using a greedy algorithm, whereby sensors are optimally placed stepwise in the most informative locations. The optimal number of sensors is obtained by evaluating the expected value of information of each of the different sensor configurations, consisting of a number and a spatial layout. The methodology is illustrated using a plate-like structure with a bounded damage area where damage can potentially arise according to a given spatial prior distribution. The results highlight the ability of the proposed methodology in rationally providing optimal sensor configurations according to a sensor cost law.