Measurements of the frequency dependence of ultrasonic backscatter are presented for specific angles of insonification for regions of infarcted and noninfarcted human myocardium. A 5-MHz transducer was used to insonify cylindrical cores taken from 7 noninfarcted regions and 12 infarcted regions of the left ventricular free wall of 6 formalin-fixed human hearts explanted because of ischemic cardiomyopathy. The dependence of apparent (uncompensated for diffraction effects and attenuation) backscatter on frequency was approximated by a power-law dependence, |B(f)|2= af(n). Under ideal conditions in a lossless medium, the effect of not compensating for the effects of diffraction and attenuation leads to the value of n to be 2.0 for Rayleigh scatterers while the frequency dependence of the fully compensated backscatter coefficient would be f4. The value of n was determined over the frequency range, 3-7 MHz. Both noninfarcted and infarcted myocardium exhibited anisotropy of the frequency dependence of backscatter, with maxima occurring at angles that were perpendicular to the predominant myofiber direction and minima when parallel to the fibers. Perpendicular insonification yielded results for n of 1.8±0.1 for noninfarcted myocardium and 1.2±0.1 for infarcted myocardium while parallel insonification yielded results of 0.4±0.1 for noninfarcted and 0.0±0.1 for infarcted myocardium. The functional form of the angle-dependent backscatter is similar for both noninfarcted and infarcted myocardium, although the frequency dependence is clearly different for both tissue states for all angles of insonification. The results of this study indicate that the anisotropy of the frequency dependence of backscatter may play a significant role in ultrasonic imaging and is an important consideration for ultrasonic tissue characterization in myocardium.