The Western Cape Province of South Africa has introduced a green economy plan called “Green is Smart”. This initiative has the envisaged possibility of providing the Province with a sustainable economy. The transition towards a green economy will, however, have implications on the food crop production in the Province. Agriculture is a vital part of the Province’s economy and a “systems thinking” approach is required to better understand how this transition will influence food crop production. The aim of this study is then to better understand systems thinking, identify different system modelling approaches, and to better understand how the Western Cape’s agriculture acts as a complex system. By achieving this, the green economy transition can be better managed within the Province’s food crop production. After reviewing the literature, system dynamics modelling was identified as the preferred modelling technique to better understand the implications of a green economy transition of the Western Cape’s food crop production. The model simulates the production for ten different food crops from 2001 until 2040. Food crops are produced with a combination of different framing practices, namely conventional, organic and conservation farming. There are three different green economy scenarios (pessimistic, realistic and optimistic), and one scenario where current practices are continued (business as usual). The model results indicate that all three green economy scenarios will require significant financial investment. The results also indicate that only the optimistic green scenario might be worth the financial investment when considering the potential benefits. The study further provides recommendations for stakeholders in order to help this transition to a green economy within the Western Cape food crop sector. The study highlights the usefulness of using system dynamics to model and better comprehend complex systems. The limitations of system dynamics modelling are also discussed in this study. Difficulties with obtaining historical data and modelling sporadic events are the two most noteworthy limitations.