It was originally our intention to study the suitability of the nitroblue-tetrazolium (NBT) test for differentiating the causes of granulocytic pleocytosis of the cerebrospinal fluid (CSF). However, several methodological problems were encountered and required extensive preliminary studies. The test was initially performed on 51 CSF samples from 37 patients using the method described by Park et al. in 1968. Although bacterial meningitis was demonstrated in 19 patients, the NBT test resulted in more than 11% NBT positive granulocytes in only four cases. No NBT positive cells were found in 24 of the samples tested. These rather poor results are not surprising if one considers that NBT is only taken up by granulocytes as macrocomplex containing heparin or fibrinogen. CSF does not contain measurable quantities of fibrinogen--either under normal conditions or in cases of viral or bacterial meningitis. A comparative study of the NBT test with and without the addition of heparin was performed using CSF samples from six patients with proven bacterial meningitis. Without heparin there were less than 11% NBT positive granulocytes in five cases, while all samples demonstrated more than 11% positive cells after addition of heparin. Best results were achieved with 70 IU of heparin in each sample. Preparation and evaluation were facilitated by use of 0.1% NBT solution and concentration of the CSF cells in a sedimentation chamber.