A radionuclide technique was developed to obtain reproducible measurements of cerebrospinal fluid flow in Rickham-Holter type shunt systems. In attempting to reproduce previously described radionuclide procedures, we found that different flow rate measurements can be obtained simply by changing the radiotracer injection technique or the method of analysis of the time activity curves which are generated. The importance of these technical pitfalls cannot be over-emphasized, since there is no gold standard to evaluate cerebrospinal fluid flow in vivo. An experimental model with a calibrated Harvard pump and a shunt system filled with normal saline was constructed. Aliquots of 500 microCi of pertechnetate in 0.05 ml of saline were injected serially into the center of Rickham reservoir using a size 28 needle. Time-activity curves were obtained after waiting five min to reach equilibrium, and standard curves of radiotracer clearance vs actual flow for each system used clinically in our institution were generated. In addition, a technique to quantify and thus correct for the small but significant leak of radiotracer which occurs at the site of injection in the reservoir is described. Excellent correlation was obtained between our results and the clinical and radiological findings in a preliminary clinical evaluation.