The paper proposes a method for modal-based damage assessment in structures where historical data on their modal properties is often missing. This method is based on the measurements of the structure's eigenfrequencies and predefined assumed damage states. Damage is defined at the level of confirming damage existence with a rough estimate of its extent, within the limits of the predefined assumptions. The factors that affect the outcomes of experimental modal analysis of damaged reinforced and prestressed beam elements on structures in use are briefly outlined and the challenges encountered during these analyses and potential solutions are briefly discussed. The method's application is demonstrated on a damaged prestressed concrete simple beam. Bayesian formulation for probability estimation is used to calculate the probability that a beam, characterized by a specific measured natural frequency, is either undamaged or has a certain level of damage. This calculation is based on the results from finite element models created in the Abaqus software suite. In the finite element models, an eigenfrequency distribution, based on the modulus of elasticity distribution, of a prestressed beam is obtained, for different levels of the assumed prestressing force drop and the corresponding damages caused by the force drop. In the presented problem, the modulus of elasticity of concrete is incorporated into the analysis as an uncertain parameter with a normal distribution. Similarly, other uncertain parameters of the actual structure can be modelled.