In this study, we investigated the use of MRI-derived thermal imaging for determining the exposure parameters for focused ultrasound (FUS) surgery. Since the temperature rise induced by a FUS beam scales linearly with power, the temperature maps acquired during subthreshold sonications can be used to determine the power necessary to produce thermal tissue damage with a desired size. Thermal images acquired during multiple sonications delivered at different locations in rabbit thigh muscle and brain tissue in vivo were analyzed to test this hypothesis. First, the linearity of the induced temperature rise with the acoustic power was tested. Next, the temperature maps acquired during preliminary low power sonications were scaled up until the estimated size of the tissue damage was equal to the tissue damage size of subsequent high power sonications. A threshold thermal dose was used to estimate the onset of thermal damage. The predicted power (based on amount of scaling required to reach the target size) was then compared to the true high power value. Overall, the temperature rise varied linearly with power (slope of deltaThigh/deltaTlow vs Power(high)/Power(low) = 0.97, 0.93 for pairs of sonications at each location in brain, muscle). The predicted power matched the true high power in the brain sonications (slope = 1.04). The predicted power underestimated the true high power in the muscle sonications (slope = 0.87). This under-prediction was due to a deviation from linearity in those cases where tissue damage was detected in subsequent MR images (slope of deltaThigh/deltaTlow vs Power(high)/Power(low) = 1.02, 0.84 for no tissue damage, tissue damage). The source of this deviation was not clear from these experiments. Even with this underestimation of the power, this method will be useful because it will allow an estimate of the proper power to use during FUS surgery without exact knowledge of the tissue parameters.