Aims: The conventional analysis of dobutamine-atropine stress echocardiography (DASE) is poorly defined and subject to considerable variation. The aim of this study was to investigate the reproducibility of strictly standardised qualitative analysis in DASE. Methods and results: Strict criteria for standardised DASE interpretation were defined through logistic regression analysis on categorical parameters obtained from 20 patients with coronary artery disease (CAD) and 20 healthy controls subjected to DASE. Three expert echocardiographers analysed DASE recordings from 100 consecutive patients referred for coronary angiography. Specificity for CAD and for predicting significant stenosis of a major coronary artery was 94% (95% CI: 83-100%) and 79% (95% CI: 63-96%), whereas sensitivity was 49% in both cases (95% CI: 38-60% and 37-61%). Within and between observer reproducibility was moderate to fair (κ = 0.56 and 0.38; 95% CI: 0.40-0.72 and 0.24-0.52). In patients without prior myocardial infarction and in echogenic patients within observer reproducibility was good (κ = 0.72 and 0.74; 95% CI: 0.52-0.92 and 0.56-0.92). Conclusions: Observer variation was not eliminated in standardised qualitative DASE interpretation based on criteria that predicted the presence of CAD with a high specificity and reproducibility was good only in certain subgroups of patients.