In Denmark, a political decision was taken to achieve, before 1997, a 50% reduction in the bulk of active ingredients in pesticides consumed, compared with average consumption in the period 1981/1985. To meet this requirement, a computer-based decision support system has been developed. In order to minimize doses, the system combines the concept of factor adjusted doses and expert knowledge. Herbicide doses are adjusted on the basis of the sensitivity of different weed species and their growth stages. Based on actual field observations, the system suggests suitable herbicides, the normal dose, the actual dose, prices per ha, etc. In future, doses will probably also be factor-adjusted according to climatic conditions, competitive ability of cereal cultivars and soil type. The need for control of individual weed species (thresholds) is based on expert knowledge. Ongoing long-term studies on seed production by weeds, competition studies etc. will hopefully help to establish more exact models for economic thresholds. After satisfactory validation through field trials, the system will be continuously updated with new findings in weed research. The system has been on the market in Denmark since spring 1991.