Neonatal CSF Reference Interval Interpreter

Neonatal CSF white cell count, protein and CSF:blood glucose ratio read against age-banded reference intervals — because the adult upper limits would call almost every healthy newborn abnormal, and the published neonatal limits do not agree with each other.

Last updated 16 Sep 2026

Neonatal CSF white cells, protein and glucose ratio

Age band + three parameters → interpretation

Age bandThe bands are the ones the source series actually published. Preterm limits come from a 318-infant neonatal intensive care series (148 preterm, 170 term); the 0-to-28-day and 29-to-60-day limits come from a 7,766-infant cohort across 18 emergency departments. Beyond 60 days the intervals continue to fall towards adult values and this page stops, because the series it is built from stop.

Was the tap blood-stained?This is asked first because it can invalidate everything else. Blood introduced at the needle carries white cells and protein into the sample, and the correction formulas in circulation (1 white cell per 500 to 700 red cells, 0.01 g/L of protein per 1000 red cells) are not agreed. The Royal Children's Hospital guidance is blunt: the safest interpretation of a traumatic tap is to count the total white cells and disregard the red cell count.

CSF white cell count (cells/µL)Cells per microlitre, which is numerically the same as ×10⁶/L — a count of 12 cells/µL is 12 ×10⁶/L. Enter the total white cell count. The neutrophil count matters separately: published 95th centiles for CSF neutrophils are 16 cells/µL in preterm and 8 in term neonates, against zero in the reference tables most UK laboratories print.

CSF total protein (g/L)Grams per litre, the form a UK report prints. Multiply by 100 for mg/dL and by 1000 for mg/L: 1.18 g/L is 118 mg/dL and 1180 mg/L. Getting this wrong by a factor of ten is the commonest entry error on a neonatal CSF, because an adult upper limit of 0.4 g/L and a neonatal one of about 1.2 g/L invite the assumption that the larger number must be a typing mistake.

CSF glucose (mmol/L)In mmol/L. Divide mg/dL by 18: the published 5th centiles of 35 mg/dL at 0 to 28 days and 37 mg/dL at 29 to 60 days are 1.9 and 2.1 mmol/L. The absolute CSF glucose is far less useful than the ratio to a blood glucose taken at the same time, which is why both are asked for.

Blood glucose at the same time (mmol/L)A blood glucose drawn at, or as close as possible to, the lumbar puncture. The ratio is what is interpreted, and a blood glucose taken an hour later — after a feed, or after intravenous glucose was started — produces a ratio that is an artefact of the timing. The ratio itself is unit-free, so mmol/L and mg/dL give the same answer provided both figures are in the same unit.

CSF:blood glucose ratio below 0.6Example

Term baby, 11 days old, clear tap: white cells 12 cells/µL, protein 1.0 g/L, CSF glucose 2.6 mmol/L with a paired blood glucose of 4.5 mmol/L

Published 95th centiles — and how far apart they are

Four series, four different term-neonate white cell limits: 10, 14, 16 and 19 cells/µL. This page uses 16 for the 0-to-28-day band because that cohort is 7,766 infants against 318 and 380, and 12 for preterm from the intensive care series, which the systematic review corroborates. A count between about 10 and 20 is inside somebody's reference interval and outside somebody else's, and that spread — not any single number — is the honest answer.
Series	Population	CSF white cells (95th centile)	CSF protein (95th centile)
7,766 infants, 18 emergency departments	0 to 28 days	16 cells/µL (median 4)	118 mg/dL = 1.18 g/L (median 66)
7,766 infants, 18 emergency departments	29 to 60 days	11 cells/µL (median 2)	91 mg/dL = 0.91 g/L (median 49)
380 infants, emergency department	28 days or less	19 cells/µL (median 3)	not reported
380 infants, emergency department	29 to 56 days	9 cells/µL (median 2)	not reported
318 infants, neonatal intensive care	Term	14 cells/µL	159 mg/dL = 1.59 g/L
318 infants, neonatal intensive care	Preterm	12 cells/µL	209 mg/dL = 2.09 g/L
Systematic review, 7 studies, 366 neonates	Term	under 10 cells/µL	110 mg/dL = 1.10 g/L
Systematic review, 7 studies, 366 neonates	Preterm	under 12 cells/µL	210 mg/dL = 2.10 g/L
Adult laboratory interval, for comparison	Adult	about 5 cells/µL	about 0.4 g/L

What a normal CSF does and does not exclude

The first two rows are why a normal cell count is not a discharge. The last four are why it is worse after antibiotics: pretreatment moves the two chemistry parameters towards normal, leaves the cell count alone, and sterilises the culture within hours. A baby given antibiotics before the tap has a CSF whose reassuring numbers were partly produced by the treatment.
Finding in culture-proven neonatal meningitis	Figure
Neonates with 0 to 1 CSF white cells	5%
Neonates with fewer than 3 CSF white cells	10%
Sensitivity of a 21 cells/µL cut-off	79% (specificity 81%)
Sensitivity of any detectable white cells	97% (specificity 11%)
Positive CSF culture with a negative blood culture	38%
Effect of 12 hours or more of antibiotics on CSF white cells	Not significantly associated
Effect of 12 hours or more of antibiotics on CSF glucose	Median 48 against 29 mg/dL
Effect of 12 hours or more of antibiotics on CSF protein	Median 121 against 178 mg/dL
CSF culture sterilisation after parenteral antibiotics	Meningococcus sterile beyond about 3 hours; pneumococcus and group B streptococcus by about 4

The parameters that do not have age bands on this page

The glucose ratio is the only parameter on this page whose threshold is age-independent, which also makes it the one least likely to be misread off an adult report. It is also the most vulnerable to a badly timed blood sample, because CSF glucose lags plasma glucose.
Parameter	What is published	Why it is not banded here
CSF:blood glucose ratio	0.6 or more in both under-1-month and over-1-month bands	No source moved the threshold in the neonate, so this page does not either
CSF neutrophil count	95th centile 16 cells/µL preterm, 8 term	Reported in the systematic review but not in the large cohorts, so it is stated rather than used as a rule
CSF lactate	0.9 to 2.5 mmol/L under 1 month; 1.0 to 2.0 over 1 month	A separate test with its own page and its own caveats
Red cell count	No interpretable interval	The guidance is to disregard it and count the total white cells, not to correct with it
15 days to 2 months, for protein and cells	Covered by the 0-to-28-day and 29-to-60-day bands above	The emergency department cohort spans it; the neonatal intensive care series does not

The reference interval is the diagnosis

Neonatal cerebrospinal fluid is the classic laboratory trap, and the trap is not subtle. An adult upper limit for CSF white cells is about 5 cells per microlitre and for protein about 0.4 g/L. The median protein in healthy term newborns is 0.66 g/L. Half of all well babies in the first month would be reported as having a raised CSF protein against an adult interval, and a count of 12 white cells — the median plus a little — would be reported as a pleocytosis when it sits comfortably inside the 95th centile of 16. The report is not wrong; the interval printed beside it is.

The awkward part is that the neonatal intervals do not agree either. Four retrievable series give 95th-centile white cell counts for a term neonate of 10, 14, 16 and 19 cells per microlitre, and protein limits of 110, 118 and 159 mg per decilitre. They differ because their populations differ: an emergency department cohort of 7,766 febrile infants is not a neonatal intensive care series of 318 babies, 72% of whom had already had antibiotics. This page uses the largest cohort for the term and older-infant bands and the intensive care series for preterm, prints all four, and says plainly that a count between about 10 and 20 is inside one published interval and outside another. Presenting a single all-ages threshold would be the real error, and it is the error the adult report makes.

Two things then limit what any of these numbers can do. The first is the tap. Blood introduced at the needle carries white cells and protein with it, and the correction formulas in circulation — one white cell per 500 to 700 red cells, 0.01 g/L of protein per 1000 red cells — are not agreed and are not endorsed by the guidance that quotes them. The safest reading is to count the total white cells and disregard the red cell count, which means a blood-stained tap cannot be used to explain away a high count and cannot be used to reassure on a normal one.

The second is that a normal CSF does not exclude neonatal meningitis. In the series that correlated cultures with parameters, 5% of neonates with culture-proven bacterial meningitis had 0 to 1 white cells and 10% had fewer than 3; the best single cut-off, 21 cells, reached 79% sensitivity. Prior antibiotics make this worse in a specific and instructive way: twelve hours or more of pretreatment raised the median CSF glucose from 29 to 48 mg/dL and lowered the median protein from 178 to 121, and left the white cell count untouched. Cultures go first of all — meningococcus sterile beyond about three hours of parenteral treatment. So in the baby who has already been treated, the two parameters that look most reassuring are the two the treatment moved, the cell count is the only one still telling the truth, and the culture may have nothing left to grow. That is when the molecular panel and the clinical picture have to carry the decision, and the reference interval, however carefully chosen, cannot.

Frequently asked questions

What is a normal CSF white cell count in a newborn?

It depends on the series, and the spread is the point. The largest cohort, 7,766 infants across 18 emergency departments, gives a 95th centile of 16 cells/µL at 0 to 28 days and 11 at 29 to 60 days, with medians of 4 and 2. A 380-infant series gives 19 and 9. A neonatal intensive care series gives upper reference limits of 14 in term and 12 in preterm infants, and a systematic review gives under 10 term and under 12 preterm. All of these are far above an adult upper limit of about 5 cells/µL.

What is a normal CSF protein in a newborn?

About 1.1 to 1.6 g/L as an upper limit in a term newborn, depending on the series, and about 2.1 g/L in a preterm infant — against an adult upper limit of roughly 0.4 g/L. The median in healthy term newborns is 0.66 g/L, which is itself above the adult limit. Protein falls with postnatal age and falls more slowly in preterm infants.

What CSF:blood glucose ratio is normal in a neonate?

0.6 or more, which is the same lower limit given for infants over one month. It is the one parameter on this page whose threshold does not move with age. The blood glucose must be drawn at or very close to the lumbar puncture, because CSF glucose lags plasma glucose and a sample taken after a feed or after intravenous glucose was started produces a falsely low ratio.

Can a traumatic tap be corrected to give a usable white cell count?

Not reliably. The correction formulas in circulation — one white cell per 500 to 700 red cells, and 0.01 g/L of protein per 1000 red cells — are not agreed, and the guidance that quotes them prefers a different approach: count the total white cells and disregard the red cell count. That cuts both ways. A raised count on a blood-stained tap cannot be explained away, and a normal count on one is not reassurance.

Does a normal CSF exclude meningitis in a neonate?

No. In the series correlating CSF cultures with CSF parameters in culture-proven neonatal meningitis, 5% of affected neonates had 0 to 1 white cells and 10% had fewer than 3, and a cut-off of 21 cells/µL achieved only 79% sensitivity. A negative blood culture does not help either: 38% of neonates with a positive CSF culture had one.

Why is a CSF less interpretable after antibiotics?

Because antibiotics move the chemistry and not the cell count. Twelve hours or more of pretreatment raised the median CSF glucose from 29 to 48 mg/dL and lowered the median protein from 178 to 121 mg/dL, while showing no significant association with the white cell count. Cultures sterilise faster still — meningococcal cultures were sterile beyond about three hours of parenteral antibiotics. So the reassuring numbers in a pretreated baby may have been produced by the treatment.

Related calculators

References

Cerebrospinal fluid reference values for young infants undergoing lumbar puncture. Pediatrics. 2018;141(3):e20173405 — 7,766 infants across 18 emergency departments; the 0-to-28-day and 29-to-60-day medians and 95th centiles used here.
Cerebrospinal fluid reference ranges in term and preterm infants in the neonatal intensive care unit. J Pediatr. 2012 — 318 infants; upper reference limits of 12 cells/µL preterm and 14 term for white cells, and 209 and 159 mg/dL for protein.
Defining cerebrospinal fluid white blood cell count reference values in neonates and young infants. Pediatrics. 2010;125(2):257 — 95th centiles of 19 cells/µL at 28 days or less and 9 at 29 to 56 days.
Normal values for cerebrospinal fluid in neonates: a systematic review. Neonatology. 2021;118(6):629 — seven studies, 270 term and 96 preterm neonates; 95th centiles for white cells, neutrophils, protein and the 5th centile for glucose.
Normative cerebrospinal fluid profiles in febrile infants. J Pediatr. 2011 — 677 atraumatic taps, and the 0-to-16/mm³ white cell reference range reported for traumatic taps.
Neonatal meningitis: what is the correlation among cerebrospinal fluid cultures, blood cultures, and cerebrospinal fluid parameters? Pediatrics. 2006;117(4):1094 — 5% with 0 to 1 white cells, 10% with fewer than 3, 79% sensitivity and 81% specificity at 21 cells/µL, and 38% with a negative blood culture.
Effect of antibiotic pretreatment on cerebrospinal fluid profiles of children with bacterial meningitis. Pediatrics. 2008 — 245 patients; 12 hours or more of pretreatment raised median CSF glucose and lowered median protein without significantly affecting the white cell count.
Royal Children's Hospital Melbourne. Clinical Practice Guideline: CSF Interpretation — age-banded normal values including the CSF:blood glucose ratio of 0.6 or more, and the recommendation to count total white cells and disregard the red cell count on a traumatic tap.

Medical Disclaimer: The tools and content provided here are for educational and reference purposes only. They are not intended to substitute for professional medical advice, diagnosis, or treatment. Clinical decisions should always be based on the comprehensive assessment of a qualified healthcare professional.