Development and use of crop models has become an interdisciplinary field with the increase on one hand of the biophysical knowledge and the development of numerous dedicated software platforms on the other hand. This trend calls for formalizing the integration of the disciplinary contributions of scientists having different expertises needed in crop modelling : agronomists, mathematicians, software scientists. We propose an interdisciplinary framework to : (1) describe the model with different expert domain knowledge, (2) identify bridges to help scientists of different domains to work together on the same model, (3) give a methodological basis (engine) for crop model studies. This theoretical framework is thus based on the integration of: (1) the agronomic domain describing the simplified functioning of soil-plant systems and the associated hypotheses; (2) the mathematical domain giving the system of equations and its properties and (3) the software domain defining the software components and their relationships. We show how these views can be specifically defined for the description and studies on crop models, how they are related to each other, and how this framework integrates some relevant concepts on integrated modelling. We finally illustrate the use of the framework for building a new crop model (a pea crop model) by changing an existing one (a wheat crop model).