In the widespread use of medical cyclotrons for isotope production, radiological and economic consequences related to the decommissioning of particle accelerators are often neglected. However, decommissioning regulation and its related procedures always demand efforts and costs that can unexpectedly impact on budgets. The magnitude of this impact depends strongly on the residual radioactivity of the accelerator and of the vault, and more specifically on the kind and activity concentration of residual radionuclides. This work reports and discusses a case study that analyzes in detail the characterization activities needed for optimized management of the decommissioning of a medical cyclotron vault. In particular, this paper presents the activities carried out for assessing the activity concentrations and for guiding the disposal of the cyclotron vault of the Italian National Cancer Institute of Milano (INT). An unshielded 17 MeV cyclotron vault was characterized by high resolution gamma-ray spectrometry both in-situ and in-laboratory on extracted samples. Monte Carlo simulations were also carried out to assess the overall distribution of activation in the vault. After a few months from the final shutdown of the accelerator, activity concentrations in the concrete walls due to neutron activation exceeded the clearance levels in many regions, especially close to the cyclotron target. Due to the relatively long half-lives of some radionuclides, a time interval of about 20 y after the end of bombardment is necessary for achieving clearance in some critical positions. Far from the target or in positions shielded by the cyclotron, activation levels were below the clearance level. The comparison between Monte Carlo simulations and experimental results shows a good agreement. The in-situ measurements, simpler and economically advantageous, cannot completely replace the destructive measurements, but they may limit the number of required samples and consequently the decommissioning costs. The methodology described and the results obtained demonstrated that it is possible to obtain accurate estimations of activity concentrations with cheap and quick in-situ measurements if the concentration profile in-depth inside the wall is well known. This profile can be obtained either experimentally or numerically through suitably validated Monte Carlo simulations.