Falsely elevated values for CSF total protein, as determined in the Beckman Astra-8 Total Protein Module, were shown to be due to a combined effect of sample cross-contamination and reaction cup contamination by serum samples on the same sample tray. Normal CSF may be misdiagnosed as having an abnormally high protein concentration, due to this interference by normal serum samples. To avoid this carryover phenomenon, it is recommended that 3 saline samples be analysed for total protein before the analysis of the CSF samples; alternatively, all the CSF samples should be analysed separately in one batch.