OBJECTIVES: To empirically assess the consequences of the inclusion of the learning curve (LC) in the HTA of medical devices by evaluating its potential impact on i) the evaluation of procedural outcomes of medical devices (MD), ii) the evaluation of rising and avoided costs; iii) the estimation of ICER; iv) the interpretation of economic evaluations in the light of planning the delivery of specific services. METHODS: A systematic review was conducted to identify the current state of knowledge about the LC in the economic evaluation of MDs and its inclusion in HTA processes. Based on the findings, the authors empirically estimated the LC for three MDs using real world data. Multivariate trend analysis assuming different parametric models both at the operator and institutional level was applied. Economic evaluation was performed with the inclusion of the LC to compare the effects. Additional analyses: comparison of LC parameters across different MDs; test of the persistence of the LC from procedural outcomes to final endpoint. RESULTS: LC effects are particularly evident for devices with high degree of complexity of use and high discontinuity with the previous technology. The learning effects are more evident on procedural outcomes and tend to gradually loose significance once final endpoints are considered. As a consequence, the LC has a more measurable impact on costs, even though in some cases also effectiveness is impacted. CONCLUSIONS: The inclusion of an observed LC could be extremely useful for HTA bodies. First, the initial “inefficiencies” could be identified, assessed, and considered as costs for the system. Second, the ICER evaluated after the learning plateau is reached could be different from the one including the learning phase, but both are needed for a fully informed HTA and for an optimal planning aimed at balancing local access and minimization of learning phase-related costs.