The conceptual modelling task in a simulation project is very important and yet is still generally regarded as more of an art than a science. The meaning and nature of conceptual modelling are discussed and a framework set out. The overall aim should be to choose the best model for the project and conceptual modelling can be viewed as a difficult optimisation problem that can be tackled effectively using a creative search process that develops alternative models and predicts their performance throughout the project. An experiment relating model characteristics to some aspects of performance is described and this type of experiment may inform the process of predicting model performance. Based on advice from the literature and my own previous work on conceptual modelling 17 principles of conceptual modelling are suggested. Conceptual modelling research is still at an early stage and ideas for future research are proposed.