What Is the Normal Range for a Ki-67 Score?

There is no single “normal range” for a Ki-67 score that applies across all tissues and diseases. Ki-67 is a protein found in cells that are actively dividing, and pathologists report it as the percentage of tumor cells that stain positive for it. In normal, non-cancerous breast tissue, fewer than 3% of cells express Ki-67. In cancer, the score can run anywhere from under 5% in slow-growing tumors to nearly 100% in the most aggressive ones, and the cutoff that matters clinically shifts depending on what type of cancer is being evaluated.

What Ki-67 Actually Measures

Ki-67 is present in cells during every active phase of division but absent in cells that are resting. A pathologist stains a tissue sample with an antibody that binds to Ki-67, then counts what fraction of tumor cells light up. The result, often called the Ki-67 labeling index or proliferation index, is expressed as a percentage. A score of 30% means roughly 30 out of every 100 tumor cells were caught in the act of dividing. Higher numbers generally signal faster-growing tumors, which tend to behave more aggressively but also sometimes respond better to chemotherapy, since chemo targets dividing cells.

In healthy tissue, Ki-67 expression is low because most cells are not actively proliferating. A study of normal breast tissue found Ki-67 in fewer than 3% of cells, and only in certain cell populations.¹ That baseline gives you a rough sense of what “normal” looks like in non-cancerous tissue, but once a tumor is present, the question becomes how high Ki-67 is relative to clinically meaningful thresholds for that particular cancer.

Breast Cancer and the 14% Threshold

The most widely discussed Ki-67 cutoff comes from breast cancer, where the score helps distinguish between two important subtypes of hormone-receptor-positive disease. Luminal A tumors, the slower-growing subtype, have been defined as estrogen-receptor-positive, HER2-negative, and Ki-67 below 14%. Luminal B tumors share the same receptor profile but have Ki-67 at or above 14%.² That distinction is not just academic: luminal B cancers tend to grow faster, carry a worse prognosis, and are more likely to benefit from adding chemotherapy to hormonal treatment.

However, the 14% cutoff has been debated since it was first proposed. Researchers have found that patients whose Ki-67 falls between 14% and 19% occupy a gray zone where prognosis is harder to pin down, and outcomes partly depend on other factors like progesterone receptor expression.³ Some expert panels have suggested using 20% or even 30% as an alternative threshold, and in practice many oncologists treat that intermediate zone with extra caution, often turning to genomic assays for a second opinion rather than relying on Ki-67 alone.

The Lab Variability Problem

One reason no single Ki-67 number has been universally adopted as a definitive cutoff is that different laboratories can report noticeably different scores on the same tumor. An international reproducibility study found that when some of the world’s most experienced labs scored the same set of centrally stained slides, their average Ki-67 values ranged from about 7% to 24%.⁴ When each lab also used its own staining protocol, the spread widened further, from about 6% to 30%. Factors driving the disagreement included which area of the tumor pathologists chose to count, whether they used formal counting methods or estimated by eye, and how they judged borderline staining.

A separate study across 30 laboratories confirmed substantial differences: the median Ki-67 labeling index on the same breast cancer cases ranged from under 1% to 33% depending on the lab, and the proportion of cancers classified as the favorable luminal A subtype varied from 17% to 57%.⁵ That kind of variability means a Ki-67 score of 15% from one lab might reflect the same biology as a score of 10% or 22% from another. It is the main reason expert guidelines have stopped short of mandating a single hard cutoff for treatment decisions in breast cancer.

Even after laboratories calibrated to a standardized scoring method and worked from centrally stained slides, clinically important discrepancies persisted in the intermediate range between 10% and 20%, exactly the zone where cutoffs are usually applied. Among cases where at least one of 22 labs reported a score in that range, none had all labs agree on a score within it.⁶ The takeaway: Ki-67 is most reliable at the extremes. A score of 5% almost certainly indicates slow proliferation, and a score of 50% clearly signals fast growth. Between about 10% and 25%, the number deserves some skepticism about its precision.

Why Tissue Handling Can Shift Your Score

Ki-67 scores are also sensitive to what happens to the tissue before anyone looks at it under a microscope. The type of fixative used, the time between surgery and fixation, and how long the tissue sits in fixative all influence how much Ki-67 staining the pathologist sees. Research has shown that insufficient fixation causes a significant drop in the Ki-67 index, and a 16-hour delay before fixation also reduces Ki-67 staining.⁷ Prolonged fixation gradually reduces Ki-67 further in a time-dependent way. Interestingly, other markers like estrogen receptor and HER2 are less vulnerable to these handling issues, which makes Ki-67 uniquely finicky among routine biomarkers.

The choice of fixative matters in cytology specimens as well. One commonly used preservative solution dramatically inhibits the standard Ki-67 antibody (MIB-1), reducing staining from a mean of 47% in formalin-fixed samples to just 10% in samples fixed with the alternative solution. Switching to a different Ki-67 antibody clone recovered the expected staining levels.⁸ If you are comparing Ki-67 results from two different biopsies or two different time points, the lab processing details matter just as much as the biology.

Ki-67 Ranges in Neuroendocrine Tumors

Outside of breast cancer, Ki-67 plays its most clearly defined clinical role in neuroendocrine tumors, where it directly determines the tumor’s grade. The World Health Organization grading system for pancreatic and gastrointestinal neuroendocrine tumors uses Ki-67 as a primary criterion: Grade 1 tumors have a Ki-67 index below 3%, Grade 2 tumors fall between 3% and 20%, and Grade 3 tumors exceed 20%. This grading is essential for both prognosis and treatment planning.⁹ A Grade 1 neuroendocrine tumor might be watched with surveillance alone, while a Grade 3 tumor usually demands aggressive treatment.

The grading boundaries in neuroendocrine tumors are among the sharpest Ki-67 cutoffs in any cancer type, partly because the clinical differences between grades are so stark. Grade 1 tumors grow slowly and are often curable with surgery. Grade 3 tumors behave much more like conventional aggressive cancers. Accurate Ki-67 scoring is especially important because these tumors sometimes look deceptively bland under the microscope even when they are proliferating rapidly.¹⁰

Lymphomas and Blood Cancers

Ki-67 behaves very differently across the spectrum of lymphomas, and the numbers reflect the enormous biological diversity of these diseases. At the indolent end, conditions like marginal-zone lymphoma and chronic lymphocytic leukemia show Ki-67 values as low as 2% to 4%.¹¹ At the aggressive end, Burkitt lymphoma can push Ki-67 close to 100%, reflecting a tumor that is dividing at an extraordinary rate. Diffuse large B-cell lymphoma typically falls somewhere in between, with Ki-67 values ranging from roughly 40% to 90%.¹²

In mantle cell lymphoma, Ki-67 serves as a prognostic separator. Patients whose tumors had Ki-67 at or below 20% survived a mean of about 70 months, compared with roughly 48 months for those above 20%.¹³ Within follicular lymphoma, the Ki-67 index climbs with grade: low-grade follicular lymphoma averages around 2%, while grade 3 disease averages close to 20%.¹¹ These numbers illustrate why a Ki-67 of 15% might be entirely expected in one lymphoma subtype and alarming in another.

Lung Cancer, Brain Tumors, and Prostate Cancer

In non-small cell lung cancer, a meta-analysis found that higher Ki-67 expression was associated with worse survival overall, with roughly a 56% higher risk of death compared to patients with lower scores.¹⁴ But the picture gets more complicated when you break it down by cell type. In lung adenocarcinoma, high Ki-67 (above about 25%) was a negative prognostic sign as expected. In squamous cell carcinoma of the lung, however, high Ki-67 (above about 50%) was paradoxically associated with better survival.¹⁵ Researchers have not reached consensus on exactly where to draw the line for lung cancer, and at least one systematic review found no agreement on Ki-67’s prognostic value across published studies when analyzed by disease stage or histological subtype.¹⁶

For meningiomas, the most common primary brain tumor, Ki-67 helps predict whether the tumor will come back after surgery. Tumors with Ki-67 below about 4% recurred after a median of nearly five years, while those with higher Ki-67 recurred in under a year.¹⁷ A meta-analysis confirmed that higher Ki-67 was significantly linked to worse overall survival and a higher risk of progression or recurrence in meningiomas.¹⁸

In prostate cancer, Ki-67 is not part of routine grading the way it is in neuroendocrine tumors, but it adds prognostic information on top of the standard Gleason score. After radical prostatectomy, patients with high Ki-67 expression were roughly three times as likely to have their cancer recur compared with those whose tumors were Ki-67-negative.¹⁹ A separate study of conservatively managed prostate cancer patients found that Ki-67 was an independent predictor of both cancer-specific and overall survival.²⁰

Tracking Treatment Response

Ki-67 is not just a one-time snapshot. In breast cancer, changes in Ki-67 during the first weeks of treatment have been used to gauge whether a therapy is working. When patients receive hormonal therapy before surgery (neoadjuvant endocrine treatment), a drop in Ki-67 after two weeks is considered a validated way to compare how well different hormonal agents are performing. Research has shown that Ki-67 levels during treatment are more closely tied to long-term outcomes than the pre-treatment score.²¹ A landmark trial found that short-term changes in Ki-67 during neoadjuvant treatment with different hormonal drugs correlated with later recurrence-free survival, suggesting that an early Ki-67 drop could someday help personalize treatment choices.²²

Can AI Fix the Scoring Problem?

Given how much Ki-67 scores vary between pathologists and laboratories, there has been growing interest in using artificial intelligence to do the counting. The results so far are promising. In one study, when pathologists of all experience levels used AI-assisted tools, their agreement with a gold standard improved substantially, with all reaching excellent concordance scores above 0.80, compared to poor or moderate consistency when junior and intermediate pathologists scored by eye alone.²³ Another study found that individual AI analysis results were confirmed by the majority of pathologists in about 96% of Ki-67 cases, and pathologists’ own scoring with versus without AI assistance agreed roughly 88% of the time on clinically relevant categories.²⁴

AI tools have an inherent advantage for this task: they do not get tired, they do not unconsciously shift their threshold partway through a slide, and they count every cell in the selected region rather than estimating. Automated Ki-67 detection has shown strong concordance with manual scoring in breast cancer.²⁵ If AI-assisted scoring becomes standard practice, it could narrow the between-lab variability enough to make fixed Ki-67 cutoffs more reliable.

Ki-67 Outside of Cancer

Ki-67 is not exclusively a cancer marker. Because it flags any dividing cell, it can also be useful in non-malignant conditions where abnormal cell proliferation is the underlying problem. Psoriasis is a good example. Research has found that Ki-67-positive cell counts in the skin are significantly higher in psoriasis compared with other skin conditions that look similar, and Ki-67 expression correlates with clinical severity.²⁶ Pathologists have used Ki-67 to help tell psoriasis apart from other inflammatory skin diseases that mimic it under the microscope, since psoriasis produces notably higher counts, especially in the upper layers of the epidermis.²⁷

Emerging Alternatives to Ki-67

Given the reproducibility issues that have dogged Ki-67 for decades, researchers have looked for alternatives that might be easier to score consistently. One candidate is phosphohistone H3, a protein that marks cells only during a very specific moment of division (mitosis). Because it stains are more sharply positive-or-negative with less ambiguity, pathologists tend to agree with each other more when counting it. Studies in breast cancer have found that observer agreement was substantially higher for phosphohistone H3 than for Ki-67, with one reporting an inter-observer concordance of 0.86 for phosphohistone H3 compared to 0.55 for Ki-67.²⁸ In one comparative study, the two markers showed a strong statistical correlation, but phosphohistone H3 was significantly faster to interpret and showed modestly higher inter-observer concordance.²⁹

Another comparative study went a step further, finding that phosphohistone H3 was the only proliferation variable that correlated with disease-free survival in their breast cancer cohort, while Ki-67 did not reach significance. The same study confirmed phosphohistone H3 was more reproducible, with a five-rater concordance score significantly higher than Ki-67’s.³⁰ Despite these advantages, phosphohistone H3 has not yet displaced Ki-67 in routine clinical practice. Ki-67 remains deeply embedded in treatment guidelines and pathology workflows, and switching to a new marker would require rebuilding the enormous body of clinical evidence linking specific thresholds to outcomes.