The invention relates to a cerebrospinal fluid shunt flow detection computing method. According to the method, since waveforms, which are formed by time sequence signal functions and time of samplingsignals when cerebrospinal fluid which carries a same exciting source signal flows into any two sampling points, are similar, and subsequently, a time-function relation of the any two sampling pointsis established, so that by calculating the maximum value of the time functions, a time difference of the any two sampling points can be obtained, and subsequently, the flow velocity and the flow rat of the cerebrospinal fluid can be obtained. The computing method provided by the invention, in comparison with existing CT and MRI detection, can improve efficiency, save money and avoid radiation; andthe computing method, in comparison with such existing computing methods as calculating extreme values in a curve, a heat conduction equation and the like, can improve the convergence and robustnessof a numerical solution, so that patient's cerebrospinal fluid shunt flow detection under any conditions can be satisfied.