Reference values for cerebrospinal fluid (CSF) pressure are 100-200 mm H2O. Elevations are due to increased intracranial pressure. The two most common causes of elevated CSF pressure are meningitis and subarachnoid hemorrhage. Brain tumor and brain abscess will cause increased intracranial pressure in most cases but only after a variable period of days or even weeks. An increase is present in many cases of lead encephalopathy. The CSF pressure varies directly with venous pressure but has no constant relationship to arterial pressure. The Queckenstedt sign makes clinical use of this information; increased venous pressure via jugular vein compression increases CSF pressure at the lumbar region, whereas a subarachnoid obstruction above the lumbar area prevents this effect.


Normal CSF is clear and colorless. It may be pink or red if many red blood cells (RBCs) are present or white and cloudy if there are many white blood cells (WBCs) or an especially high protein content. Usually there must be more than 400 WBCs/mm3 (0.4 Ч 109/L) before the CSF becomes cloudy. When blood has been present in the CSF for more than 4 hours (literature range, 2-48 hours), xanthochromia (yellow color) may occur due to hemoglobin pigment from lysed RBCs. Protein levels of more than 150 mg/100 ml (1.5g/L) may produce a faint yellowish color that can simulate xanthochromia of RBC origin. Severe jaundice may also produce coloration that simulates xanthochromia.


Reference values are 45 mg/100 ml (2.5 mmol/L) or higher (“true glucose” methods). Values of 40-45 mg/100 ml are equivocal, although in normal persons it is rare to find values below 45 mg/100 ml. The CSF glucose level is about 60% of the serum glucose value (literature range, 50%-80%). In newborns, the usual CSF level is about 80% of the serum glucose level. It takes between 0.5 and 2 hours for maximum change to occur in CSF values after a change in serum values.

The most important change in the CSF glucose level is a decrease. The classic etiologies for CSF glucose decrease are meningitis from standard bacteria, tuberculosis, and fungi. Occasionally in very early infection the initial CSF glucose value may be normal, although later it begins to decrease. A frequent impression from textbooks is that one should expect a decreased CSF glucose value in nearly all patients with acute bacterial meningitis. Actually, studies have shown that only 60%-80% of children with acute bacterial meningitis have CSF glucose levels below the reference range. Studies on patients of various ages have shown decreased glucose levels in 50%-90% of cases. A major problem is the effect of blood glucose levels on CSF glucose levels. Since elevated blood glucose levels may mask a decrease in CSF values, it is helpful to determine the blood glucose level at the same time that the CSF specimen is obtained, especially in diabetics or if intravenous (IV) glucose therapy is being given. On the other hand, a low CSF glucose level may be due to peripheral blood hypoglycemia, especially if the CSF cell count is normal. Other conditions that may produce a decrease in CSF glucose level are extensive involvement of meninges by metastatic carcinoma and in some cases of subarachnoid hemorrhage, probably due to release of glycolytic enzymes from the RBCs. In leptospiral meningitis and in primary amebic meningoencephalitis, the CSF glucose level is decreased in some patients but not in others. In most other central nervous system (CNS) diseases, including viral meningitis, encephalitis, brain abscess, syphilis, and brain tumor, CSF glucose levels typically are normal when bacterial infection of the meninges is not present. However, decreased levels are sometimes found in aseptic meningitis or in meningoencephalitis due to mumps, enteroviruses, lymphocytic choriomeningitis, and in both herpes simplex virus types 1 and 2 CNS infections. For example, one investigator reports that 10% of cases of meningitis due to enterovirus had decreased CSF glucose. The box summarizes glucose level patterns.


The normal protein concentration of CSF is usually considered to be 15-45 mg/100 ml (0.15-0.45 g/L) (literature range, 9-90 mg/100 ml; 0.09-0.9 g/L). There is considerable discrepancy in the literature regarding the upper reference limit, which is most often listed as 40, 45, and 50 mg/100 ml (international system of units [SI] 0.4, 0.45, 0.5 g/L). Newborn values are different and even more uncertain. From birth to day 30, the range is 75-150 mg/100 ml (0.75-1.50 g/L), with the literature range being 20-200 mg/100 ml and the upper limit varying from 140-200 mg/100 ml. For days 30-90, the reference range is 20-100 mg/100 ml (0.2-1.0 g/L). From day 90 to 6 months of age, the reference range is 15-50 mg/100 ml (0.15-0.50 g/L). Values slowly decline, reaching adult levels by 6 months of age. Over age 60 years, some persons increase the upper reference limit to 60 mg/100 ml (0.6 g/L). As a general but not invariable rule, an increased protein concentration is roughly proportional to the degree of leukocytosis in the CSF. The protein concentration is also increased by the presence of blood. There are, however, certain diseases in which a mild to moderate protein concentration increase may be seen with relatively slight leukocytosis; these include cerebral trauma, brain or spinal cord tumor, brain abscess, cerebral infarct or hemorrhage (CVA), CNS sarcoidosis, systemic lupus, lead encephalopathy, uremia, myxedema, multiple sclerosis (MS), variable numbers of hereditary neuropathy cases, and chronic CNS infections. Diabetics with peripheral neuropathy frequently have elevated CSF protein levels without known cause. Blood in the CSF introduces approximately 1 mg of protein/1,000 RBCs. However, when the RBCs begin to lyse, the protein level may appear disproportionate to the number of RBCs. In acute bacterial meningitis, the CSF protein is elevated in about 94% of cases (literature range, 74%-99%).

A marked protein elevation without a corresponding CSF cell increase is known as “albuminocytologic dissociation.” This has usually been associated with the Guillain-Barrй syndrome (acute idiopathic polyneuritis) or with temporal (giant cell) arteritis. Actually, about 20% of patients with the Guillain-Barrй syndrome have normal CSF protein levels, and less than 25% have CSF protein levels of 200 mg/100 ml (2.0 g/L) or more.

Protein may be measured in the laboratory quantitatively by any of several methods. A popular semiquantitative bedside method is Pandy’s test, in which CSF is added to a few drops of saturated phenol agent. This agent reacts with all protein, but apparently much more with globulin. Chronic infections or similar conditions such as (tertiary) syphilis or MS tend to accentuate globulin elevation and thus may give positive Pandy test results even though the total CSF protein level may not be greatly increased. Contamination by blood will often give false positive test results.

The technical method used can influence results. The three most common methods are sulfosalicylic acid, anazolene sodium (Coomassie Blue dye), and trichloracetic acid. Sulfosalicylic acid and Coomassie Blue are more influenced by the ratio of albumin to globulin than is trichloracetic acid.

In some CNS diseases there is a disproportionate increase in gamma-globulin levels compared with albumin or total protein levels. Several investigators have noted that increased CSF gamma-globulin levels occur in approximately 70%-85% (literature range, 50%-88%) of patients with MS, whereas total protein levels are elevated in only about 25% of the patients (range, 13%-34%). Various other acute and chronic diseases of the brain may elevate CSF gamma-globulin levels, and this fraction may also be affected by serum hyperglobulinemia. The latter artifact may be excluded by comparing gamma-globulin quantity in serum and CSF.

The colloidal gold test also depends on changes in CSF globulins and at one time was considered very helpful in the diagnosis of MS and syphilitic tabes dorsalis. However, the colloidal gold procedure has very poor sensitivity (only 25% of MS patients display the classic first zone pattern) and poor specificity and is considered obsolete. When measurement of CSF protein fractions is ordered, the current standard procedure is some method of immunoglobulin quantitation.

Cell count

Normally, CSF contains up to five cells/mm3, almost all of which are lymphocytes. In newborns, the reference limits are 0-30 cells/cu mm, with the majority being segmented neutrophils. Also, one study reported at least one segmented neutrophil in 32 percent of patients without CNS disease when centrifuged CSF sediment was examined microscopically. There was correlation with elevated peripheral blood WBC count and presence of some RBCs in the sediment, presumably from minimal blood contamination of the lumbar puncture specimen not grossly evident. As a general rule, any conditions that affect the meninges will cause CSF leukocytosis; the degree of leukocytosis will depend on the type of irritation, its duration, and its intensity. Usually, the highest WBC counts are found in severe acute meningeal infections. The classic variety is the acute bacterial infection. It is important to remember that in a few cases if the patient is seen very early, leukocytosis may be minimal or even absent, just as the CSF glucose level may be normal. Although most investigators state or imply that 100% of patients with acute bacterial infection have elevated cell counts on initial lumbar puncture, in one study, 3% of patients had a WBC count less than 6 WBCs/mm3, and there have been several reports of other isolated cases. Usually, in a few hours a repeat lumbar puncture reveals steadily increasing WBC counts. Therefore, normal initial counts are not frequent but may occur and can be very misleading. Another general rule is that in bacterial infections, polymorphonuclear neutrophils usually are the predominating cell type; whereas in viral infections, chronic nervous system diseases, and tertiary syphilis, lymphocytes or mononuclears ordinarily predominate. However, there are important exceptions. One study reported that about one third of patients with usual bacterial pathogens but with CSF WBC counts less than 1,000/mm3 (1.0 x 109/L) had a predominance of lymphocytes on initial lumbar puncture (the only patient with a WBC count over 1,000 who had lymphocytosis had a Listeria infection). Another exception is tuberculous meningitis, which is both a bacterial and a chronic type of infection. In this case, the cells are predominantly lymphocytes (although frequently combined with some increase in neutrophils). A third exception is unusual nonviral organisms, such as Listeria monocytogenes (most often seen in neonates or in elderly persons or those who are immunocompromised), fungus, and spirochetes (leptospirosis and syphilis). Listeria meningitis may have lymphocytic predominance in some cases and neutrophilic predominance in others. Similarly, in active CNS syphilis the few studies available indicate that if the WBC count is elevated, lymphocytes predominate in 60%-80% of cases. On the other hand, coxsackievirus and echovirus infections may have a predominance of neutrophils in the early stages; in most of these patients, the CSF subsequently shifts to lymphocytosis. Uremia is said to produce a mild lymphocytosis in about 25% of patients. Partial treatment of bacterial meningitis may cause a shift in cell type toward lymphocytosis.

After therapy is started, WBC values usually decrease. However, in some cases more than 50 WBCs/mm3 may persist at least 48 hours and sometimes longer than 2 weeks following adequate therapy. This tends to be more common with Haemophilus influenzae infection but may also occur with pneumococcal infection. Fungal infections are more commonly associated with persistently elevated neutrophils than bacterial infections, even though fungi typically have CSF lymphocytosis rather than neutrophilia. Nocardia meningitis or brain abscess, however, is one bacterial infection that does tend to show persistent neutrophilia more often than other bacteria.

In cases of subarachnoid hemorrhage or traumatic spinal fluid taps, approximately 1 WBC is added to every 700 RBCs (literature range, 1 WBC/500-1,000 RBCs). This disagreement in values makes formulas unreliable that attempt to differentiate traumatic tap artifact from true WBC increase. Also, the presence of subarachnoid blood itself may sometimes cause meningeal irritation, producing a mild to moderate increase in polymorphonuclear leukocytes after several hours that occasionally may be greater than 500 WBCs/ mm3. Occasionally a similar phenomenon occurs in patients with intracerebral hematoma. Another exception is the so-called aseptic meningeal reaction that is secondary either to a nearby infection or sometimes to acute localized brain destruction. In these cases, which are not common, there may be a wide range of WBC values, with neutrophils often predominating. When this occurs, however, the CSF glucose level should be normal, since the meninges are not directly infected. Aseptic meningitis is not the same as aseptic meningeal reaction. Aseptic meningitis is due to direct involvement of the meninges by nonbacterial organisms. Viruses cause most cases, but organisms such as amebae sometimes may cause meningitis. Bacterial CSF cultures are negative. The CSF glucose level is usually normal. Protein concentration is usually but not always increased. The WBC count is elevated to a varying degree; the predominant type of cell depends on the etiology. True aseptic meningitis also has to be differentiated from bacterial organisms that do not grow on ordinary culture media (anaerobes, leptospira, mycobacteria, bacteria inhibited by previous or current antibiotic therapy, etc.).

Neonatal CSF reference range differences from childhood and adult values

Neonates have higher CSF reference ranges for protein, glucose, and cell count than adults have. Protein was discussed earlier. Cell counts 1-7 days after birth average about 5-20/mm3 (range, 0-32 mm3), with about 60% being segmented neutrophils. Glucose is about 75%-80% of the blood glucose level.


The diagnosis of acute bacterial meningitis often depends on the isolation of the organisms in the spinal fluid. In children, there is some regularity of the types of infection most commonly found. In infants under the age of 1-2 months, group B streptococci are most frequent, closely followed by Escherichia coli. Listeria monocytogenes is often listed as third, followed by other enteric gram-negative bacteria. In children from age 3 months to 5 or 6 years, H. influenzae is the most common organism; Meningococcus is second, and Pneumococcus is third. In older children and adolescents, Meningococcus is first and Pneumococcus is second. In adults, Meningococcus and Pneumococcus are still dominant, but Pneumococcus is more prevalent in some reports. Staphylococci are the etiology in about 4%-7% of cases, most often associated with CNS operations (e.g., shunt procedures), septicemia, or endocarditis. In old age, pneumococci generally are more common pathogens than meningococci, with gram-negative bacilli in third place. However, there can be infection by nearly any type of organism, including Listeria and fungi. Also, in patients who are debilitated or have underlying serious diseases such as leukemia or carcinoma, Listeria and fungi are not uncommon. In many cases, a centrifuged spinal fluid sediment can be smeared and Gram stained so that the organisms can be seen. A (bacterial) culture should be done in all cases where bacterial meningitis is suspected or is even remotely possible. Special provision should be made for spinal fluid to reach the laboratory as quickly as possible; and if any particular organism is suspected, the laboratory should be informed so that special media can be used if necessary. For example, meningococci grow best in a high carbon dioxide atmosphere, and H. influenzae should be planted on media provided with a Staphylococcus streak. Culture is said to detect about 85% (range, 54%-100%) of cases. In one series there was thought to be about 5% false positive culture results due to contamination. Previous or concurrent antibiotic therapy is reported to decrease culture detection rates by about 30% and Gram stain detection rates by about 20%.

Patterns of Cerebrospinal Fluid Abnormality: Cell Type and Glucose Level

Acute bacterial meningitis

Some cases of early phase acute bacterial meningitis
Primary amebic (Naegleria species) meningoencephalitis
Early phase Leptospira meningitis

Brain abscess
Early phase coxsackievirus and echovirus meningitis
CNS syphilis (some patients)
Acute bacterial meningitis with IV glucose therapy
Listeria (about 20% of cases)

Tuberculosis meningitis
Cryptococcal (Torula) meningitis
Mumps meningoencephalitis (some cases)
Meningeal carcinomatosis (some cases)
Meningeal sarcoidosis (some cases)
Listeria (about 15% of cases)

Viral meningitis
Viral encephalitis
Postinfectious encephalitis
Lead encephalopathy
CNS syphilis (majority of patients)
Brain tumor (occasionally)
Leptospira meningitis (after the early phase)
Listeria (about 15% of cases)

Gram stain

Gram stain of the sediment from a centrifuged CSF specimen should be performed in all cases of suspected meningitis. Gram stain yields about 70% positive results (literature range, 50%-90%) in culture-proved acute bacterial meningitis cases. There is some controversy over whether Gram staining need be done if the CSF cell count is normal. In the great majority of such cases bacteriologic findings are normal. However, occasional cases have been reported of bacterial meningitis without an elevated cell count, most commonly in debilitated or immunosuppressed patients. It may also occur very early in the initial stages of the disease. Gram stain yields unquestioned benefits in terms of early diagnosis and assistance in choice of therapy. However, a negative Gram stain result does not rule out acute bacterial meningitis, and some false positive results occur. The majority of false positive results are due to misinterpretation of precipitated stain or debris on the slide. In some cases the presence of bacteria is correctly reported, but the type of organism may be incorrectly identified due to overdecolorization of the organisms or insufficient familiarity with morphologic variations of bacteria.

Latex agglutination tests for bacterial antigens

Recently, rapid slide latex agglutination (LA) tests have become available for detection of pneumococcal, meningococcal, H. influenzae type B, and streptococcal group B bacterial antigen in CSF or in (concentrated) urine. Until the 1980s, CSF bacterial antigen detection was done by counterimmunoelectrophoresis (CIE). This method had an overall sensitivity of about 60%-70% (range, 32%-94%). The LA kits are considerably faster, easier, and have increased detection rates (about 85%-90%; range, 60%-100%). There is some variation in overall sensitivity between different manufacturers’ kits. This is especially true for antibodies against meningococci. The major reason is that several different strains of meningococci may produce infection, and it is necessary to have an antibody against each one that it is desired to detect. The most common strains are type B (about 47% of cases; range 28%-50%), type C (about 30%; range, 23%-63%), type Y (about 10%), type W135 (about 10%) and type A (about 3%). Some kits do not include antibodies against all of these strains. Besides not reaching 100% sensitivity, the LA kits are relatively expensive per patient (in part due to the multiple tests included in each kit to detect the different organisms). Besides LA, there is a commercially available kit based on a coagglutination method. This kit has an overall sensitivity about intermediate between CIE and LA, with similar results to LA for H. influenzae, and less sensitivity for pneumococci and meningococci.

Cerebrospinal fluid lactate

A number of published reports have evaluated CSF lactic acid assay in various diseases. In general, patients with acute bacterial, tuberculous, and fungal meningitis have elevated CSF lactate values, whereas normal persons and those with viral (“aseptic”) meningitis do not. Sensitivity of the test for acute bacterial meningitis is about 93%-95% (range, 66%-100%). Most reports indicate clear-cut separation between bacterial and viral infection values (in the sense that the values in viral meningitis are not elevated), but several investigators found that some patients with viral meningitis have elevated CSF lactate levels (about 20% of cases in these reports). However, in all instances of viral meningitis the elevation was less than twice the reference range upper limits. One advantage of CSF lactate is that it may remain elevated 2-3 days after the start of antibiotic therapy. However, values in some cases of treated or partially treated bacterial infection do return to normal relatively early.

Xanthochromia has been reported to nonspecifically increase CSF lactate levels, although one report states that blood itself does not. CNS tissue destruction from various causes (including brain tumor, head trauma, CVAs, and intracerebral hemorrhage, cerebral hypoxia, and seizures), may also produce elevated CSF lactate levels.

Because CSF lactate is not specific for bacterial infection, because it is not elevated in all cases of bacterial meningitis, and because there is some uncertainty as to whether lactate elevation excludes the possibility of viral meningitis, the true usefulness of the test is not clear. The availability of LA slide tests further complicates the picture. However, CSF lactate assay could be useful in patients with symptoms of meningitis if CSF Gram stain results are negative and LA test results (if available) are also negative. Such patients could have tuberculous or fungal meningitis, partially treated bacterial meningitis, or meningitis due to other organisms. Increased CSF lactic acid levels, especially if more than twice the upper reference limit, could suggest that further investigation is essential. A normal lactate level is not reliable in excluding bacterial meningitis.

Reference range for CSF lactate is usually based on specimens from children or adults. One investigator found that neonates 0-2 days old had mean values nearly 60% higher than those obtained after 10 days of life, whereas neonates aged 2-10 days had about 25% higher levels than after age 10 days.

Computerized tomography, magnetic resonance imaging, and radionuclide brain scanning

Computerized tomography (CT) and magnetic resonance imaging (MRI) are now important aids in screening for abnormality in the CNS. Both can visualize the ventricular system as well as detect mass lesions both within CNS tissue and outside. In addition, the nature of the lesion can frequently be deduced from density characteristics. Both CT and MRI technology have been changing rapidly, and assessment of their capabilities relates to data currently available. Detection of brain tumors is to some extent influenced by location and type of tumor as well as by technical factors such as use of IV contrast media. Reports indicate CT abnormality in approximately 90%-95% of patients with mass lesions or areas of tissue destruction. Statistics from CT and MRI are not always comparable to radionuclide brain scans, since data from the scans vary according to the number of head positions employed, the isotope preparation used, the time between administration of isotope and scanning, and whether a blood flow study was included. In general, CT is about 10%-15% more sensitive than brain scan in cerebral tumor or about 5%-10% more sensitive when brain scanning is performed with optimal technique. It is somewhat more reliable than brain scanning in detecting posterior fossa lesions. MRI is about 5% more sensitive than CT. In chronic subdural hematoma, detection with CT is about equal to that achieved when brain scanning is combined with cerebral blood flow study. In acute subdural or epidural hematoma, CT is significantly better than radionuclide brain scanning. An important advantage over radionuclide techniques in either acute or chronic subdural hematoma is that CT can frequently permit a more exact diagnosis, whereas abnormalities found by radionuclide techniques are often not specific. MRI is reported to be equal to or slightly better than CT. In cerebral infarct, all three techniques are affected by the time interval after onset. During the first week, CT is somewhat more sensitive than MRI or radionuclide scanning without blood flow study but still detects only 50%-60% of infarcts. The sensitivity of all techniques increases to the 80% range by 3-4 weeks. Intracerebral hematoma is much better seen on CT than brain scan regardless of the time interval and better than MRI in the early stages. None of the three techniques is perfect. In most of the various focal lesion categories, a certain percentage are detected by one technique but not the other, although CT and MRI are generally superior to radionuclide scans. When Hakim’s syndrome (normal pressure hydrocephalus) is a possibility, CT or MRI can rule out the disorder by displaying normal ventricular size. If ventricular dilation is seen, cerebral atrophy may be inferred in some cases, but in many the differentiation between atrophy and normal pressure hydrocephalus cannot be made with adequate certainty.

Overall advantages of CT over standard radionuclide procedures are ability to visualize the ventricular system, a relatively small but definite increase in detection rate for brain tumors, more specificity in the appearance of many lesions, and better delineation of CNS anatomy. Advantages over MRI are lower cost, better results in early CVA or early hemorrhage, and ability to detect calcifications and show details of bone (which is seen poorly on MRI). Advantages of radionuclide procedures are elimination of the need for x-ray contrast media (which many CT patients must receive), lower cost for equipment and lower charge to the patient, and ability to inspect major blood vessels via blood flow studies. Advantages of MRI are absence of any radiation, slightly and sometimes significantly better detection rate of many lesions compared to CT, better tissue detail, much better visualization of the spinal cord, and in some cases ability to suggest a more exact diagnosis. The major disadvantages are considerably higher cost, slower imaging time, and in some cases, problems with patients who have cardiac pacemakers or internal metal objects.