The potential for supplementing sparse economic data and aggregate production data with a detailed specification of land quality constraints obtained from a Geographical Information System, in order to parameterise a regional model of rice acreage response in Vietnam, is investigated in this paper. It is asserted that such an approach has advantages in terms of improved credibility amongst agricultural scientists cum policy makers because of the detailed attention to land capability, compared to conventional elasticity based approaches that are commonly used for agricultural policy analysis when data is scarce. The approach relies on the use of Positive Mathematical Programming to calibrate the acreage response for land classes where more than one farming system is observed in the baseline data. Sensitivity of the model to the nature of the calibration assumptions are presented, and results regarding acreage response with respect to rice prices, in different seasons and regions are presented and discussed. It is concluded that while the current model relies on some heroic assumptions regarding interpretation of the available information, it could be improved considerably, and at low cost, if taken up and developed within the Ministry. There remains, however, a research agenda regarding farm household decision making in rice based production systems of Vietnam that could provide for improved calibration in the future.