The world population is expected to grow to 10 billion people in 2040, requiring food production at least to triple, as a combined result of population growth and dietary changes. In this study scenarios for food security are developed for 15 major regions of the world. Food production is calculated on a 1°x1° grid basis, using inputs from a digital data base containing soil, climatic, agronomic and demographic data. This allows us to study impacts of different production scenarios and their environmental side effects. Soil area data are derived from the digitized FAO soil map, considered to be representative for soil units on a 1°x1° grid; weather variables of 978 stations are allocated to the grid cells. Food production estimations are made with a simple crop growth module, a soil water balance and a soil nitrogen balance. Calculations are done on a grid basis, each grid cell being characterized by its suitability for arable farming or grassland, soil and climatic conditions and the availability of irrigation water. Results of Yield Oriented Agriculture (YOA) and Environment Oriented Agriculture (EOA) production scenarios on food self-sufficiency are presented. If we assume that there is no limit to transport of food across the globe, YOA allows the entire globe an affluent diet, while EOA allows only a moderate diet. For this scenario, regional self sufficiency indices vary widely: most regions can produce food required for an affluent diet, but not East, South and West Asia that account for 48% of the expected global population in 2040. Also Southeast Asia (9% of expected global population in 2040) and West and North Africa (10%) come close to the lower limit for food self-sufficiency. With EOA, only the former USSR, North and South America, Central and Southern Africa and Oceania can offer their future populations an affluent diet. Southern Asia will experience food shortages even under minimum food demand, and less affluent diets do not provide a solution. In this paper, attention is given to the methodology regarding soils, and to the uncertainty regarding soil and water in input data. Particularly soil suitability for agriculture for each of the grid cells has a large effect on model calculations, as it is a rather rough approximation. Knowledge of soils across the world shows important weaknesses with respect to the extent of coverage, extrapolation of point observations to grid cells, definition of soil characteristics compatible with crop models, and handling of preferential flow in soil profiles. We emphasize the need for digitized, reliable and readily available natural resource information in the fields of soil surveys, climatology, irrigation water availability, land use and land cover. The ultimate goal of the approach is not to predict the future of global agriculture, but to allow decision makers to compare various scenarios for food self-sufficiency, and to raise awareness regarding issues related to food security. In spite of the above-mentioned shortcomings in basic data, it is demonstrated that it is possible to develop the required scenarios now, instead of having to wait on more reliable soil information.