The diagnosis of coronary artery disease (CAD) on the basis of inducible ischemia in > or = 2, rather than 1, segments was suggested to improve specificity of dobutamine stress echocardiography (DSE). However, the impact of using these criteria on the sensitivity and accuracy of DSE was not studied. We studied the accuracy of DSE (up to 40 microg/kg/min) for the diagnosis of CAD in 290 patients with suspected myocardial ischemia using the criteria of > or = 1 and > or = 2 ischemic segments. Ischemia was defined as new or worsening wall motion abnormalities using a 16-segment model. Among the 85 patients without previous myocardial infarction, significant CAD was detected in 51 (60%). Sensitivity, specificity, and accuracy of DSE using > or = 1 ischemic segment were 73%, 85%, and 78%, respectively. Those using > or = 2 segments were 67%, 94%, and 78%, respectively (p = NS). Regional specificity improved by using > or = 2 segments (91% vs 96%, p <0.05) at the expense of an equivalent reduction in regional sensitivity (60% vs 44%, p <0.05), whereas the regional accuracy was similar (80% vs 79% for > or = 1 and > or = 2 segments, respectively). In the 205 patients with previous myocardial infarction, the criterion of ischemia in > or = 1 segment had a higher sensitivity and accuracy for overall diagnosis of CAD (75% vs 64%, p <0.05; 77% vs 68%, p <0.05), infarct-related CAD (64% vs 47%, p <0.005; 70% vs 57%, p <0.01), and remote CAD (74% vs 57%, p <0.005; 78% vs 69%, p <0.05) than the criterion of > or = 2 segments, respectively. It is concluded that in patients without previous myocardial infarction, the use of > or = 2 ischemic segments by DSE for the diagnosis of CAD does not improve the accuracy of DSE compared with the criterion of > or = 1 ischemic segment. Conversely, in patients with previous infarction the use of > or = 2 segments reduces the overall and regional sensitivity and accuracy without a significant improvement in specificity.