Despite the increasing sophistication and quality of published work, the development of a cumulative body of knowledge and an evidence-base for Information Systems (IS) research still represents a major challenge. IS research is still predominantly undertaken by IS researchers for other IS researchers and not utilized to its full extent by IS practitioners or policy makers. We focus on this problem and express the need for a new evidence-based research perspective. It is argued that it is time to refocus the efforts of IS academics (and practitioners) to develop a new evidence-base for IS research whereby it can more routinely inform, develop, improve and support IS practice. We contribute to this debate by defining evidence-based practice (EBP), its relevance to IS, and the need to develop an evidence-based approach. We look in particular at its brief history, and its subsequent evolution, development and widespread acceptance in Medicine; making reference to recent arguments and critiques of EBP in other disciplines such as software engineering and management. We espouse the need to develop a similar evidence-based movement and infrastructure within the IS research and practitioner communities and then put forward a possible roadmap for the development of Evidence-Based Information Systems (EBIS) that comprises 9 key initiatives. We conclude our argument by stating that the current extent, severity and impact of IS failures are unacceptable, emphasizing the need for a new perspective for IS research that encourages and incorporates EBP as a guiding principle to inform better IS practice.