Bottom-up models of oil production are continuously being used to guide investments and policymaking. Compared to simpler top-down models, bottom-up models have a number of advantages due to their modularity, flexibility and concreteness. The purposes of this paper is to identify the crucial modeling challenges, compare the different ways in which nine existing models handle them, assess the appropriateness of these models, and point to possibilities of further development. The conclusions are that the high level of detail in bottom-up models is of questionable value for predictive accuracy, but of great value for identifying areas of uncertainty and new research questions. There is a potential for improved qualitative insights through systematic sensitivity analysis. This potential is at present largely unrealized. (C) 2013 Elsevier Ltd. All rights reserved.