In recent years, different types of bio-economic models have been developed to support the analysis of the potential impact of agrarian policies on changes in land use, sustainable resource management and farmers' welfare. Most bio-economic models rely on series of discrete input–output coefficients for current and improved cropping and livestock activities, whereas mathematical programming procedures are usually applied to analyse optimum allocative choice. Adequate procedures for the smooth integration of biophysical information into economic decision models are, however, not readily available. This article provides a new and comprehensive framework for the incorporation of technical input–output coefficients derived from agroecological simulation approaches into bio-economic farm household models. Therefore, continuous production functions are estimated for the production side of the farm household model, making use of meta-modelling principles. It is shown that meta-modelling offers considerable scope for improving the specification and behaviour of bio-economic farm household models. This procedure is applied in a farm household model developed for the analysis of farmers’ response to agrarian policies in Southern Mali. Results are presented for the behaviour of a typical household, focusing attention on the trade-offs between farm income and soil nutrient balances under free market conditions and with constraints on labour, capital and animal traction markets. The stability and robustness of the model is analysed through a simulation of the impact of higher input costs for land use and fertiliser applications.