Purpose: Decision aids (DAs) aim to inform medical decision making by providing patients with data on the benefits and risks of medical interventions, while also acknowledging their personal preferences. DAs have generally been found to improve the quality of medical decisions, as indicated by greater patient knowledge and satisfaction with the decision process. Some have argued that DAs may also reduce spending on expensive health care interventions, but the evidence concerning this hypothesis has not been reviewed systematically. We systematically reviewed clinical trials of decision aids to characterize their impact on patients' intent to use and actual utilization of medical interventions that could affect costs. Methods: The Medline-PUBMED database (1966-¬2007) was searched for all randomized controlled trials using DAs. The search was supplemented by a review of the 2003 Cochrane Review on Decision Aids (1966-12/2002). All references in these papers were also reviewed. We required that studies report information on patients' intent to use specific treatments or actual treatment utilization. Where cost data was not collected, the effects of the DA on costs were inferred using estimates of the cost of the interventions whose utilization was affected by the DA. We did not include the cost of the DA or its administration. Results: Thirty six RCTs were identified and 32 provided clear information on the proportion of patients choosing alternative health interventions. Most studies concerned DAs for risk-screening and risk-management rather than treatments. The most common forms of DAs were paper-based (15 studies) and video-based (13 studies). Other forms of DAs included audio booklets, computers, and discussion. The effect of the DA on costs ranged from a decrease in cost of $7000 per patient to an increase in costs of $1500 per patient. The use of DAs resulted in patients opting for the less expensive medical intervention in 56% of studies and the more expensive one in 44%, but the majority of DAs affected costs by less than $50 per patient). Conclusions: Decision aids varied greatly in their effects on utilization and costs with most having little effects on costs though there was some slight tendency for DAs to more often reduce than increase costs. Future research should consider whether the potential effects of DAs on costs affects when DAs are developed and used.