The process of crop modelling to develop operational software requires different skills, from conceptualization of the biophysical system to computer programming, involving three main scientific disciplines: agronomy, mathematics, and software engineering. Model building implies transforming a conceptual model into sets of mathematical equations and then translating these equations into a computer program. Although recent crop modelling frameworks can technically support model building, the modelling process is not always well documented and difficult to repeat. The focus of this paper is therefore on developing and documenting an approach to re-assemble crop models, i.e. develop a new model from an existing one, using a crop modelling framework and crop physiological knowledge. Modifications to an initial crop model were classified according to three categories: (i) changes in parameter values, (ii) changes in equations, and (iii) changes in overall model structure. We illustrate the approach with a case study transforming a wheat crop model into a pea crop model. We discuss the role of each actor in the process to document diverse uncertainties related to the model (i.e. contextual situation, data, structure), and the general applicability of the approach for different crop modelling frameworks. We conclude that the use of our approach to re-assemble a crop model within a modelling framework facilitates integration of different disciplines around a modelling objective, and facilitates creating transparent and reproducible models. (Résumé d'auteur)