The objective of this work was to determine changes in the ultrasound properties of heated tissues, with potential application to monitoring of minimally invasive thermal therapy (MITT). Changes in backscatter coefficients and frequency-dependent attenuation coefficients were measured over the frequency range 2.5 MHz to 5 MHz from heated samples of store-bought fresh bovine liver, which was used as a tissue model. Individual liver samples were heated from 37Β°C to either 5O"C, 55"C, 6O"C, 65Β°C or 70Β°C by warm water. The backscatter coefficient increased during the first 3 min by a factor of 1.09 and 1.11 before the tissue reached 50Β°C and 55"C, respectively. A decrease in backscatter coefficient followed at 50Β°C by a factor of 1.12 below the initial level and, at 55"C, the backscatter coefficient dropped below the initial level by a factor of 1.19. The backscatter coefficient decreased within the first 2 mln by a factor of 1.22 before the tissue reached 6O"C, then increased gradually to a factor of 1.05 below the initial level. At 65Β°C and 7O"C, the changes in backscatter coefficient were highly variable, which may have been due to production of gas microbubbles in the heated tissues. The ultrasound attenuation coefficient increased by as much as 1.48 dB cm-' over a 30-mln period at 70Β°C. First-order rate parameters derived from the attenuation results revealed one rate process at 50Β°C and 55Β°C and two rate processes at 6O"C, 65Β°C and 70Β°C. An activation energy of 1.00 x lo4 cal mol-' was derived from the second rate constants at 6O"C, 65Β°C and 7O"C, which indicates that changes in attenuation may be due to protein denaturation. In conclusion, ultrasound image monitoring of thermal therapy treatment in liver may be feasible; however, the backscatter coefficient changes during heating are small and are of the same order as the variation in these changes from point to point in the tissue. 0 1997 World Federation for Ultrasound in Medicine & Biology.