Background In 2000 the Korean government initiated efforts to secure healthcare accessibility and efficiency anytime and anywhere via the nationwide healthcare information system by the end of 2010. According to the master plan, electronic health record (EHR) research and development projects were designed in 2005. One subproject was the design and implementation of standards-based interoperable clinical decision support (CDS) capabilities in the context of the EHR system. Objective The purpose of this study was to describe the challenges, process, and outcomes of defining and implementing a national CDS architecture to stimulate and motivate the widespread adoption of CDS services in Korea. Methods CDS requirements and design principles were established by conducting a selective literature review and a survey of clinicians, managers, and hospital and industrial health information technology engineers regarding issues related to CDS architectures. The previous relevant works of the American Medical Informatics Association, the Healthcare Information and Management Systems Society, and Health Level Seven were used to validate the scope and themes of the service architecture. The Arden Syntax, Standards-Based Sharable Active Guideline Environment, First DataBank, and SEBASTIAN approaches were used to assess the coverage of the application architecture thus defined. A CDS prototype of an outpatient hypertension management system was implemented and assessed in a simulated experimental setting to evaluate the feasibility of the proposed architecture. Results Four CDS service features were identified: knowledge application, knowledge management, audit and evaluation, and CDS and knowledge governance. Five core components of CDS application architecture were also identified: knowledge-execution component, knowledge-authoring component, data-interface component, knowledge repository, and service-interface component. The coverage and characteristics of the architecture identified herein were found to be comparable with those described previously. Two scenarios of deployment architecture were identified in the context of Korean healthcare. The preliminary feasibility test revealed that the architecture exhibited good performance and made it easy to integrate patient data. Conclusion We have described the efforts that have been made to realize CDS service features, core components, application, and deployment architectures in the context of the Korean EHR. These outcomes showed the potential to contribute to the adoption of CDS at the national level.