Abstract: Doppler Ultrasound techniques have been used for many years to measure the flow rate of fluids. One of the challenges in using these techniques is to separate fluid flow signal from background noise. The present invention is directed to methods and apparatus for automatically distinguishing fluid flow signal from noise. It is found that noise in Doppler ultrasound techniques has characteristic of a white Gaussian spectral distribution and the mean noise can be characterized as a constant. The present invention involves measuring ultrasonic signals, generating sets of discrete spectral components from these signals, and statistically determining a maximum likelihood estimation from the spectral components. This estimation is used to calculate the fluid velocity.