Aim: To evaluate the impact of decision support on the proportion of troponin I (cTnI) tests and associated costs over the period 2000-7 for patients presenting with chest pain in an emergency department (ED) setting. Methods: A longitudinal study using linked data for patients presenting with chest pain from the ED and laboratory information systems of a metropolitan teaching hospital in Melbourne, Australia. The study period was divided into a pre-intervention period (2000-2), which contained no decision support; an initial post period (2003-4) after the introduction of a quality improvement initiative (utilising a paper-based guideline, education, audit and feedback) about cTnI test ordering and the incorporation of the guideline as a decision support feature of the computerised provider order entry system; followed by a post-modification period (2005-7) after the electronic decision support feature was modified to allow clinicians to bypass viewing the complete guideline. Results: There was a significant fall in the proportion of cTnI tests ordered per patient presentation across the three periodsdpre (2000-2), post (2003-4) and postmodification (2005-7)dfrom 7.3% to 4.1% and 2.8%, respectively. Analysis of costs showed significant reductions in the mean costs for cTnI tests per patient presentation from $A9.28 to $A8.54 and $A8.18, respectively, which amounted to a modest saving of $A13 251 since the initiation of decision support in 2003. Conclusions: Decision support systems are often part of multifaceted implementations undertaken over time. They require continuous monitoring and modifications to ensure optimal performance.