The goal of this work was to determine whether the frequency dependence of apparent backscatter coefficient (not corrected for attenuation within the myocardium) could differentiate completed, remote infarction from acute myocardial injury in vivo. Myocardial infarcts were produced in six dogs by coronary artery occlusion. One to 12 months later, acute ischemic injury was induced in each dog by ligation of a coronary artery that supplied a region of myocardium adjacent to the established infarct. Infarct, ischemic, and normal regions were interrogated with a 5-MHz, circular, 0.5-in. diam, broadband, focused, piezoelectric transducer mounted in a water-filled stand-off device placed against the exposed, beating heart. Apparent backscatter coefficients were measured over the range of frequencies from 3-7 MHz. The frequency dependence was obtained from the slope of log apparent backscatter coefficient versus log frequency. No significant difference in frequency dependence was found between normal and acutely ischemic myocardium for periods of up to 2 h of ischemia. In contrast, frequency dependence in regions of remote infarct (1.8 + 0.1, mean + standard error) was significantly lower than that in acutely ischemic or nonischemic regions (2.3 + 0.1) (p < 0.01). These results suggest that remote myocardial infarction can be differentiated from acutely injured but still potentially salvageable myocardium in vivo on the basis of the frequency dependence of backscatter.