There is a lively debate between 'concerned' and 'unconcerned' analysts regarding the future availability and affordability of oil. We critically examine two interrelated and seemingly plausible arguments for an unconcerned view: (1) there is a growing amount of remaining reserves: (2) there is a large amount of oil with a relatively low average production cost. These statements are unconvincing on both theoretical and empirical grounds. Oil availability is about flows rather than stocks, and average cost is not relevant in the determination of price and output. We subsequently implement a bottom-up model of regional oil production with micro-foundations in both natural science and economics. An oil producer optimizes net present value under the constraints of reservoir dynamics, technological capacity and economic circumstances. Optimal production profiles for different reservoir drives and economic scenarios are derived. The field model is then combined with a discovery model of random sampling from a lognormal field size-frequency distribution. Regional discovery and production scenarios are generated. Our approach does not rely on the simple assumptions of top-down models such as the Hubbert curve - however it leads to the same qualitative result that production peaks when a substantial fraction of the recoverable resource remains in-ground. (C) 2011 Elsevier Ltd. All rights reserved.