Abstract A parametric model for the ultrasound signals from blood and tissue is developed and a new imaging method, knowledge-based imaging, is defined. This method utilizes the likelihood ratio function to classify blood and tissue signals. The method separates blood and tissue signals by the difference in movement patterns in addition to the difference in powers. The prior information about the levels of expected system white noise and clutter noise are utilized to enhance the image quality. The implementation of knowledge-based imaging is outlined, and some knowledge-based images with different parameter settings are visually compared with a second-harmonic image, a fundamental image and a bandwidth image. In order to understand the parameter estimation process a computer simulation is introduced to outline the differences between the imaging methods. The apparent error rates are calculated in any reasonable tissue to blood signal ratio, tissue to white noise ratio and clutter to white noise ratio. A discussion of further development of knowledge-based imaging is also described in this paper.