The use of carbon dioxide (CO2) for enhanced oil recovery (EOR) is being tested for oil fields in the Illinois Basin, USA. While this technology has shown promise for improving oil production, it has raised some issues about the safety of CO2 injection and storage. The Midwest Geological Sequestration Consortium (MGSC) organized a Monitoring, Verification, and Accounting (MVA) team to develop and deploy monitoring programs at three EOR sites in Illinois, Indiana, and Kentucky, USA. MVA goals include establishing baseline conditions to evaluate potential impacts from CO2 injection, demonstrating that project activities are protective of human health and the environment, and providing an accurate accounting of stored CO2. This paper focuses on the use of MVA techniques in monitoring a small CO2 leak from a supply line at an EOR facility under real-world conditions. The ability of shallow monitoring techniques to detect and quantify a CO2 leak under real-world conditions has been largely unproven. In July of 2009, a leak in the pipe supplying pressurized CO2 to an injection well was observed at an MGSC EOR site located in west-central Kentucky. Carbon dioxide was escaping from the supply pipe located approximately 1 m underground. The leak was discovered visually by site personnel and injection was halted immediately. At its largest extent, the hole created by the leak was approximately 1.9 m long by 1.7 m wide and 0.7 m deep in the land surface. This circumstance provided an excellent opportunity to evaluate the performance of several monitoring techniques including soil CO2 flux measurements, portable infrared gas analysis, thermal infrared imagery, and aerial hyperspectral imagery. Valuable experience was gained during this effort. Lessons learned included determining 1) hyperspectral imagery was not effective in detecting this relatively small, short-term CO2 leak, 2) even though injection was halted, the leak remained dynamic and presented a safety risk concern during monitoring activities and, 3) the atmospheric and soil monitoring techniques used were relatively cost-effective, easily and rapidly deployable, and required minimal manpower to set up and maintain for short-term assessments. However, characterization of CO2 distribution near the land surface resulting from a dynamic leak with widely variable concentrations and fluxes was challenging. ?? 2011 Published by Elsevier Ltd.