This paper reports on a practical investigation into methodology confidence of detection (COD) in acoustic emission (AE) testing. The developed technique relies on a commercially available software technique called “source cluster analysis” that examines the number of detected signals over a specific area. Two factors that control cluster analysis are cluster size (the area that signals are detected within) and cluster threshold (the number of detected events required to trigger a cluster). A confidence of detection matrix was developed based on altering cluster size and cluster threshold which was then applied to a practical investigation of a four-point bend test monitored using AE. Fracture length in the specimen was monitored using a foil crack gauge. The varying sizes and thresholds of the confidence matrix were used in a cluster analysis of the recorded AE data, as the initial cluster formed in the fracture region a crack length could be identified (based on the foil crack gauge). Results showed that it was possible to detect a crack length of 0.55 mm with a very high level of confidence.