Abstract Background A policy of intra-operative transcranial Doppler (TCD) and completion angioscopy was previously associated with virtual abolition of intra-operative stroke (apparent upon recovery from anaesthesia) following carotid endarterectomy (CEA). The aims of this study were to determine whether the prevalence of technical error has diminished with experience and whether our monitoring/quality control policy was still associated with low rates of intra-operative stroke 20 years after its introduction. Methods Retrospective review of four consecutive cohorts of 400 patients undergoing CEA between October 1995 and March 2010 (1600 CEAS in total). Results One hundred four patients (7%) had thrombus removed following angioscopy and prior to flow restoration, while 31 (2.1%) underwent repair of a distal intimal flap. The prevalence of intimal flaps diminished from 4.9% in the first 400 patients to 0.8% in the last 400 patients ( p = 0.006). By contrast, the prevalence of retained thrombus did not decline with experience (8.5%, 3.7%, 10.3% and 5.4% for the four consecutive periods). Intra-operative TCD and completion angioscopy was, however, associated with extremely low rates of intra-operative stroke (0.25%, 0.25%, 0.5% and 0.25% during the four study periods). Conclusion Most intra-operative strokes probably follow embolisation of thrombus following restoration of flow. This can be prevented by angioscopy which has the advantage of being performed prior to flow restoration. Increasing experience was associated with a decline in the detection of intimal flaps, but not in the prevalence of retained thrombus. Even the most experienced of surgeons can still be responsible for inadvertent technical error.