To assess the effects on health care resource utilization of a network of microcomputer workstations for writing all inpatient orders. Randomized controlled clinical trial. Inpatient internal medicine service of an urban public hospital. A total of 5219 internal medicine patients and the 68 teams of house officers, medical students, and faculty internists who cared for them. Microcomputer workstations, linked to a comprehensive electronic medical record system, for writing all inpatient orders. Total inpatient charges for each admission and charges for specific categories of orders. A time-motion study of selected interns assessed the ordering system's time consumption. Intervention teams generated charges that were $887 (12.7%) lower per admission than did control teams (P = .02). Significant reductions (P < .05) were demonstrated separately for bed charges, diagnostic test charges, and drug charges. Reductions of similar proportion and statistical significance were found for hospital costs. The mean length of stay was 0.89 day shorter for intervention resident teams (P = .11). Interns in the intervention group spent an average of 33 minutes longer (5.5 minutes per patient) during a 10-hour observation period writing orders than did interns in the control group (P < .0001). A network of microcomputer workstations for writing all inpatient orders significantly lowered patient charges and hospital costs. This would amount to savings of more than $3 million in charges annually for this hospital's medicine service and potentially tens of billions of dollars nationwide. However, the system required more physician time than did the paper charts. Research at other sites and system advances to reduce time requirements are warranted.