OBJECTIVES: Today, we observe quite a disruption over most countries in value appreciation of new drugs, that is making historical data used for forecasting pharmaceutical expenditure poorly reliable. Forecasting methods rarely addressed uncertainty. The objective of this project was to propose a methodology to perform pharmaceutical expenditure forecast integrating expected policy changes and uncertainty (developed for the European Commission: “EU Pharmaceutical expenditure forecast” http://ec.europa.eu/health/healthcare/key_documents/index_en.htm METHODS: 1) Identify all pharmaceuticals going off-patent and new branded medicinal products, over 5-year forecasting period, in seven EU Member States; 2) Develop a model to estimate direct and indirect impact (based on health policy and clinical experts) on savings of generics and biosimilars. Inputs were originator sales value, patent expiry date, time to launch after marketing authorization, price discount, penetration rate, time to peak sales, impact on brand price; 3) Develop a model for new entrants, which estimated sales progression in  competitive environment. Clinical expected benefits as well as commercial potential were assessed for each product by clinical experts. Inputs were development phase, marketing authorization dates, orphan condition, market size, competitors; 4) Analyze separately budget impact of products going off-patent and new entrants according to several perspectives, distribution chain and outcomes; and 5) Address uncertainty surrounding estimations via deterministic and probabilistic sensitivity analysis.   RESULTS: This methodology allowed 1) to identify main parameters impacting pharmaceutical expenditure forecast varying across countries:  generics discounts and penetration, brand price after patent loss, reimbursement rate, penetration of biosimilars and discount price,  distribution chain, time to peak sales for new entrants; 2) to define probability of occurrence of a pre-determined impact thanks to probabilistic analysis; and 3) to test different pricing and reimbursement policy decisions on health expenditures. CONCLUSIONS: This methodology is independent of historical data and appears to be highly flexible and adapted to test robustness and provide probablisitic analysis to support policy decision making.