The design space of a pharmaceutical product can be described through mathematical models. As a result of process changes (e.g., due to fouling, clogging, environmental conditions, effect of upstream/downstream units), the prediction fidelity of a model may change during plant operation, thus questioning the appropriateness of the design space described by means of the model. In this study we propose a continuous model adaptation strategy that uses information from the running operation to quantify the extent at which a process change is impacting on the design space, hence on the possibility to meet the assigned quality target product profile at the current operating conditions and input materials properties. By combining the predictions of a first-principles model with measurements from plant sensors, the methodology jointly exploits dynamic state estimation and feasibility analysis to adapt the model-based description of the design space as the operation progresses. Namely, the state estimator is deployed to adapt the model by adjusting in real-time a subset of its parameters; feasibility analysis is then used to update the design space in real-time by exploiting the up-to-date model returned by the state estimator. The effectiveness of the proposed methodology is illustrated through two simulated case studies: the roller compaction of microcrystalline cellulose and the fermentation of penicillin in a pilot scale bioreactor. The methodology can complement a continued process verification activity within a process validation program. (C) 2019 Elsevier Ltd. All rights reserved.