What Is the Normal Range for a CA 19-9 Test?

Most laboratories flag a CA 19-9 result as normal when it falls below 37 U/mL, but that single cutoff hides real variation. A study of healthy adults found the upper limit of normal closer to 26 U/mL when age and sex were accounted for, and women consistently run higher than men. What counts as “normal” also shifts depending on the assay your lab uses, your blood type, and whether you have any underlying liver or bile duct issues. Understanding where the number comes from and what can push it around is more useful than memorizing one threshold.

Where the 37 U/mL Cutoff Comes From

The 37 U/mL figure you see on most lab reports was established decades ago by the original test manufacturers and has stuck around largely by convention. It was chosen to balance sensitivity and specificity for detecting pancreatic cancer, and most major commercial platforms still use it or something close. One widely used system (Abbott’s Alinity i) recommends 37 U/mL, while another (Roche’s COBAS e411) uses 34 U/mL.¹ That small gap matters more than it sounds: if your hospital switches assay platforms, the same blood sample can produce a meaningfully different number.

Research looking specifically at healthy people suggests the true upper boundary may be lower. In a large study of apparently healthy Chinese adults aged 20 to 60, the overall 97.5th-percentile reference interval topped out at about 26.5 U/mL. Males ranged from roughly 2 to 25 U/mL, while females ranged from about 2.4 to 29 U/mL.² That means a woman with a reading of 30 U/mL is technically within her sex-specific range but might get flagged as elevated on a report using the generic 37 U/mL cutoff. In practice, most doctors treat any result under 37 as reassuring, but the sex difference is worth knowing if your result is borderline.

Why Women Tend to Have Higher Levels

Across multiple studies and assay platforms, healthy women show higher average CA 19-9 levels than healthy men. One comparison using two different immunoassay methods in a Thai population found that women averaged roughly 10 to 14 U/mL while men averaged about 7 to 8 U/mL, depending on the platform.¹ The Chinese reference-interval study confirmed the pattern: the upper reference limit for women (about 29 U/mL) was higher than for men (about 25 U/mL).²

The biological reason is not entirely settled, but CA 19-9 is a carbohydrate antigen expressed on the surface of cells lining the digestive tract, bile ducts, and parts of the reproductive system. Tissues in the uterus and ovaries produce it as well, which likely contributes to the sex gap. Gynecological conditions, including benign ovarian cysts, can push levels far beyond the normal range even when no cancer is present.³ If you are a woman with a mildly elevated CA 19-9 and no digestive symptoms, that context matters for how your doctor interprets the result.

The Lewis Blood Group Problem

This is the single biggest blind spot of the CA 19-9 test, and it catches many people off guard. Your body can only produce CA 19-9 if you carry certain forms of the Lewis and secretor genes. Roughly 5 to 10 percent of people of European descent (and a higher proportion in some other populations) have a Lewis-negative blood type, and these individuals produce little to no CA 19-9 regardless of what is happening in their body. Their levels will read near zero even if they have advanced pancreatic cancer.

Current evidence shows that CA 19-9 is expressed mainly in people with Lewis phenotypes Le(a+b−) or Le(a−b+), while it is largely absent in those who are Le(a−b−).⁴ If your doctor orders CA 19-9 and the result is undetectably low, Lewis-negative status should be considered before anyone concludes that cancer has been ruled out. Some oncology centers now check Lewis antigen status before relying on CA 19-9 for monitoring, though this is not yet routine everywhere.

Benign Conditions That Push CA 19-9 Up

A high CA 19-9 does not automatically mean cancer. A range of non-malignant conditions can raise the number, sometimes dramatically. An older study examining patients with elevated CA 19-9 who did not have cancer found that half of them had liver cirrhosis, cholecystitis, pancreatitis, or hepatitis.⁵ Bile duct obstruction is an especially common culprit: when bile cannot flow normally, CA 19-9 backs up into the bloodstream and levels can spike into the hundreds or thousands.

Research on patients with obstructive jaundice illustrates this vividly. In people with benign bile duct blockages who underwent drainage, the average CA 19-9 level dropped by about 76 percent afterward, compared with only about 39 percent in patients whose obstruction was caused by a tumor.⁶ That difference in how fast levels fall after draining the blockage can itself be a useful diagnostic clue. A CA 19-9 that plummets once bile flow is restored usually points away from cancer.

Beyond the liver and bile ducts, conditions as varied as ovarian cysts, chronic kidney disease, thyroid problems, and even certain lung conditions have been reported to elevate CA 19-9. In people with autosomal dominant polycystic kidney disease, CA 19-9 levels correlated with liver cyst volume and a liver enzyme called gamma-GT, both independent predictors of the elevation.⁷ None of these elevations have anything to do with cancer, but they can create real anxiety when they show up on a lab report.

CA 19-9 and Pancreatic Cancer

The test was originally developed as a marker for pancreatic cancer, and that remains its primary clinical use. In people who already have symptoms suggestive of pancreatic cancer, CA 19-9 has a sensitivity of about 79 to 81 percent and a specificity of about 82 to 90 percent.⁸ In plain terms, it correctly identifies roughly four out of five people who have the disease and correctly clears roughly eight or nine out of ten who do not.

Those numbers sound decent until you consider screening, where the math falls apart. Because pancreatic cancer is rare in the general population, even a fairly specific test will produce far more false alarms than true detections. The positive predictive value of CA 19-9 as a screening tool in asymptomatic people is less than 1 percent.⁸ That means out of every 100 or more healthy people who test positive, fewer than one would actually have the disease. This is why no major guideline recommends CA 19-9 as a screening test for the general public.

An early comparison study found that raising the diagnostic cutoff from 37 to 75 U/mL improved the combined sensitivity and specificity for detecting pancreatic cancer, pushing sensitivity to about 87 percent with specificity around 93 percent. Sensitivity was higher in advanced disease (about 91 percent) than in tumors that could still be surgically removed (about 79 percent).⁹ In other words, the marker performs best precisely when you need it least, when the cancer has already spread far enough to make a cure unlikely.

Monitoring Treatment and Watching for Recurrence

Where CA 19-9 really earns its keep is not in initial diagnosis but in tracking how a known pancreatic cancer responds to treatment over time. Because the molecule has a relatively short half-life in the blood, changes in the level reflect what the tumor is doing in something close to real time, often picking up progression before imaging shows visible growth.¹⁰ A falling CA 19-9 after chemotherapy or surgery generally signals a good response; a rising one is a red flag.

For detecting recurrence after surgery, researchers have explored using the rate of rise rather than a fixed cutoff. One study found that a CA 19-9 level that rose to 1.35 times its post-surgery nadir detected recurrence with about 72 percent sensitivity and 62 percent specificity. A steeper threshold of 2.45 times the nadir traded sensitivity for better specificity (about 85 percent), which is often what clinicians want when the next step would be more surgery or aggressive treatment.¹¹ Trends matter more than any single number.

Prognostic Clues Before and After Surgery

Beyond monitoring, the absolute level of CA 19-9 before surgery can hint at how advanced the tumor is and what a patient’s odds look like. Patients with node-negative pancreatic cancer had a median preoperative CA 19-9 of just 9 U/mL, compared with 164 U/mL in those with positive lymph nodes. Earlier-stage tumors also ran lower than more locally advanced ones (41 versus 162 U/mL).¹²

At the extremes, the numbers become quite stark. A large analysis found that resectability (whether the tumor could be successfully removed) dropped from 80 percent when CA 19-9 was below 37 U/mL to just 38 percent when levels exceeded 4,000 U/mL. Five-year survival went from 27 percent down to zero across the same range.¹³ For patients who received chemotherapy before surgery, a CA 19-9 that remained above 200 U/mL afterward predicted very poor outcomes, with a two-year survival rate of only about 41 percent.¹⁴

None of this means a single CA 19-9 reading should drive the surgical decision on its own. Imaging, biopsy findings, and a patient’s overall fitness all factor in. But a very high pre-treatment CA 19-9 is a conversation-starter about expectations and treatment planning.

Uses Beyond Pancreatic Cancer

Although CA 19-9 is most tightly linked to pancreatic cancer, it is also used in the workup of bile duct cancer (cholangiocarcinoma), where it serves as an adjunct diagnostic marker. The literature has broadened its potential relevance even further. Elevated levels have been reported in colorectal cancer, esophageal cancer, and cancers outside the digestive tract, including ovarian, lung, and thyroid malignancies.¹⁵ In most of these settings, CA 19-9 is not sensitive or specific enough to be used alone, but it adds a piece to the diagnostic puzzle when combined with imaging and other markers.

Assay Variations and the Hook Effect

Not all CA 19-9 tests are created equal, and this has practical consequences. As noted earlier, different commercial platforms can produce modestly different numbers from the same blood sample. In one comparison, the mean CA 19-9 in healthy subjects was about 11.6 U/mL on one system versus about 9.4 U/mL on another, a statistically significant gap.¹ The 97.5th percentile cutoff (roughly the upper limit of normal) was about 60 U/mL on one platform and about 38 U/mL on the other. If you are tracking CA 19-9 over time, sticking with the same lab and the same assay method matters.

There is also a technical quirk called the “hook effect” that can catch labs off guard. When CA 19-9 concentrations are extremely high, the excess antigen can actually saturate the test’s antibodies and paradoxically produce a falsely low result. Some older assay kits were more vulnerable to this than others. Extended-range kits and competitive-binding assay designs have largely mitigated the problem, but laboratories still sometimes need to dilute and re-run a sample when the clinical picture does not match a surprisingly low reading.¹⁶ In short, a falsely low CA 19-9 due to the hook effect is uncommon in modern labs but not impossible, particularly when dealing with very advanced cancers that produce enormous quantities of the antigen.

Combining CA 19-9 With Other Biomarkers

Because CA 19-9 alone has clear blind spots, researchers have been working on combining it with other markers to improve detection of early pancreatic cancer. One of the more promising approaches pairs CA 19-9 with methylated DNA markers (MDMs), fragments of tumor DNA that carry distinctive chemical tags. A study measuring three such markers in plasma alongside CA 19-9 achieved an area under the curve of 0.99 in the training set and 0.90 in an independent test set of early-stage pancreatic cancer patients, significantly outperforming either marker type alone.¹⁷

A separate multicenter study took this idea further by measuring methylated DNA markers in pancreatic juice (fluid collected during an endoscopy) and combining those with plasma CA 19-9. The combination reached an area under the curve of 0.95, better than pancreatic juice markers alone (0.87) or plasma CA 19-9 alone (0.91).¹⁸ None of these combination panels are in routine clinical use yet, but they point toward a future where CA 19-9 is one ingredient in a more accurate multi-marker test rather than a standalone number.

CA 19-9 in Newborns and Children

If you are reading up on CA 19-9 because your child had bloodwork done, know that the adult reference ranges do not apply. In newborns, CA 19-9 levels are naturally higher than in adults and follow a developmental pattern: concentrations are highest at earlier gestational ages (from about 24 to 37 weeks) and fall toward term and through infancy.¹⁹ This means a premature baby’s CA 19-9 can be well above the adult upper limit without anything being wrong. Pediatric oncologists factor in gestational and chronological age when interpreting these results.

The elevated neonatal levels gradually taper as the child matures, but age-specific reference ranges in children remain less well defined than in adults. If your pediatrician orders this test, the result should be interpreted alongside the child’s age and clinical context rather than compared to the 37 U/mL adult cutoff.

When Your Doctor Orders the Test and What to Expect

CA 19-9 is a simple blood draw with no special preparation. Results typically come back within a few days. Your doctor might order it in several scenarios: as part of a workup for abdominal symptoms when pancreatic or bile duct cancer is suspected, as a baseline measurement before starting treatment for a known cancer, or as a serial measurement during chemotherapy or after surgery to track response. It is not recommended as a routine screening test for people without symptoms.

If you receive an isolated mildly elevated result, the most likely explanations are benign. Liver inflammation, gallstones, a flare of pancreatitis, or even a recent respiratory infection can nudge CA 19-9 above 37 U/mL. Your doctor will usually recheck the level after treating the underlying condition. A level that normalizes on repeat testing is reassuring. A level that keeps climbing, especially into the hundreds or thousands without an obvious benign explanation, warrants imaging and further investigation. The trend over time almost always tells a more useful story than a snapshot on a single day.