An evaluation of the validity in infants of standard computations for the correction of white cell counts in blood-contaminated cerebrospinal fluid (traumatic lumbar puncture) was undertaken. In 39 infants with significant blood contamination (greater than 10(4) red blood cells/microL) and no evidence of bacterial meningitis, both calculated corrections based on simultaneously obtained complete blood counts and estimated corrections based on average values resulted in overcorrection in a majority of cases (33 of 39 and 25 of 39, respectively). The mechanism of this overcorrection could not be defined. In eight infants with significant blood contamination and proven bacterial meningitis, correction computations normalized or overcorrected the white cell count in one case. It appears that standard computations frequently overcorrect white cell counts in blood-contaminated cerebrospinal fluid and that the magnitude of the overcorrection may obscure pathologic states in some cases.