Failure or malfunction of complex engineered networks involves relevant social and economic aspects, so that their maintenance is of primary importance. In assessing the reliability of such networks, it should be duly considered that they are a whole made of different parts, and that some of these individual parts or structures are often crucial to assure the proper operation of the entire network. Moreover, each of these structures can be considered a complex system by itself: structural reliability theory should be thus combined with advanced numerical analysis tools in order to obtain realistic estimates of the probability of failure. Accurate estimations are especially required in seismic zones, aiming to efficiently plan future interventions. This paper presents a method for the reliability assessment of a critical element of engineered networks. The method is discussed with special reference to a relevant case study: a concrete water tank, which is a key component of a water supply system. Special attention is devoted to the reliability assessment of the tank under seismic loads, that requires the definition of a Finite Element model. The model is identified on probabilistic basis, applying the Bayes Theorem and response surface methods. The proposed approach allows to significantly speed up the structural identification process, leading to sounder estimate of the input parameters. Finally, the seismic fragility curves of the structure are developed according to the relevant limit states, demonstrating that information regarding the global structural behavior and local checks can be effectively combined in structural reliability assessments.