The paper discusses various approaches to modelling measures of the Common Agricultural Policy (CAP) within a computable general equilibrium (CGE) framework for the new budgetary period 2014–2020. The task of modelling such a complex policy as the CAP with the use of CGE is not easy at least for three reasons. First, the policy itself is very complex – Pillar 2 alone includes about 17 very heterogeneous measures, which differ in terms of implementation and eligibility criteria. Pillar 1 measures are not targeted (in terms of goods and services that may be bought with these funds) and thus the assessment of their impact requires additional knowledge on how they were spent. Second, although CGE models represent all sectors of the economy, yet they normally do not characterise individual sectors with such a precision as would be desired for modelling the nuances of the individual CAP measures. Third, the CAP evolves towards less tangible measures (risk management, quality improvements, conditionality based on environmental requirements), and towards increasing role of non-marketed goods (provision of public goods, environmental amenities, food safety). There is also an increasing role of human capital manifested by e.g. bottom-up approaches or co-operation measures. They, however, are difficult to grasp by the CGE models since they are not directly observed or linked to the exogenous variables controlled in this types of models. While taking all the challenges into account and relaying on a literature review the article presents some solution and makes suggestions for possible ways of modelling new CAP 2014–2020 within CGE modelling framework which may be useful in the policy evaluation.