DUR is a process of problem detection and intervention designed to improve the quality and economy of drug prescribing. Retrospective DUR attempts to detect and address patterns of prescribing that might be indicative of inappropriate therapy. When the process is extended to a largely ambulatory population such as Medicaid beneficiaries, a number of complications are introduced due to the large numbers of patients and sparsity of data. In order to examine the impact of implementing a Medicaid DUR program, we developed a system that would apply screening criteria to prescription claims. It has been used to screen prescribing of groups of two antihypertensive drugs in the 1990 Maryland Medicaid population for 177,409 Medicaid eligible individuals. Potentially significant problems were detected with respect to dosing, duplication of therapeutic agents and drug interactions. The system represents, we believe, a significant improvement in the ability to detect and report prescribing decisions by increasing the specificity of the detection system. By the application of this system to a set of real-world data, we have demonstrated that it is feasible to implement such a system and derive results that are potentially useful in reducing the incidence of inappropriate physician decision-making.