The laser Doppler technique is used to assess tissue perfusion. Traditionally an integrated, ω-weighted (firstorder filter) power spectrum is used to estimate perfusion. In order to be able to obtain selective information about the flow in vessels with different blood cell velocities, higher order filters have been implemented, investigated, and evaluated. Theoretical considerations show that the output of the signal processor will depend on the flow speed, for a given concentration of blood cells, according to Sout∝νn where υ is the average blood cell speed and n is the spectral filter order. An implementation of filters using zero-, first-, second-, and third-order spectral moments was utilized to experimentally verify the theory by using a laser Doppler perfusion imager. Two different flow models were utilized: A Plexiglas model was used to demonstrate the various signatures of the power spectrum for different flow speeds and filter orders, whereas a Delrin model was used to study the relationship between the flow velocity and the output of the signal processor for the different filters. The results show good agreement with theory and also good reproducibility. Recordings made on the skin of the wrist area demonstrated that the flow in small veins can be visualized by the use of higher spectral orders.