During a six-month interval, 1,677 cerebrospinal fluid (CSF) specimens with cell counts of 0 to 5 x 106/L were examined by one of six physician microscopists. Malignant cells were found in ten of these CSF specimens. In seven, unidentified cells had been noted on prior review by a technologist. Technologists had a frequency of false negative errors of 0.12%. Physicians had a frequency of apparent false positive errors of 0.12%. There is little or no benefit from physician microscopic examination of all CSF specimens that have normal cell counts. Adoption of a different policy would reduce medical care costs by $40,000 for this entire series of 1,677 specimens without adversely affecting the quality of laboratory results.