The BI-RADS Scale: What Each Category Means

BI-RADS is a standardized scoring system that radiologists use to classify breast imaging findings on a scale from 0 to 6, with each number carrying a specific meaning about what was seen and what should happen next. Developed by the American College of Radiology starting in the late 1980s, the system was created to solve a real problem: radiologists across different hospitals were describing the same mammogram findings in wildly different ways, leaving referring physicians and patients confused about what a report actually meant. Today, every mammogram, breast ultrasound, or breast MRI you receive will include a BI-RADS category, and understanding what yours means can save you a lot of unnecessary worry or, in some cases, prompt you to act quickly.

Why the System Was Created

Before BI-RADS existed, a radiologist in one city might describe a finding as “likely benign” while another called the same thing “indeterminate,” and neither term carried a clear recommendation for what to do next. The ACR launched the BI-RADS initiative to standardize both the vocabulary radiologists use to describe what they see and the management recommendations tied to each assessment category.¹ The system gives radiologists a shared dictionary of descriptors for specific imaging features, along with a reporting structure that links each assessment to a recommended course of action.² It also provides a framework for tracking outcomes over time, so institutions can audit whether their assessments match what biopsies and follow-ups eventually reveal.

Categories 0, 1, and 2

These three categories sit at the low end of the scale and, with one exception, are reassuring results.

Category 0 means the imaging study is incomplete. The radiologist saw something that needs more information before a final assessment can be made. This might mean you need additional mammogram views, a spot compression, or an ultrasound to get a clearer look. Category 0 is not a diagnosis; it is a request for more pictures. If you receive a BI-RADS 0 after a screening mammogram, follow-up diagnostic imaging is recommended within 60 days.³ Many people find a Category 0 result alarming, but it is common and usually resolves to a benign finding once the additional imaging is done. Dense breast tissue, overlapping structures, or a technical issue with the original images are frequent reasons for the callback.

Category 1 is a negative result. Nothing abnormal was found. There are no masses, suspicious calcifications, or areas of architectural distortion. You return to routine screening on whatever schedule your age and risk profile call for.

Category 2 is also a benign finding, but the radiologist is noting something specific that is clearly not cancer. This could be a simple cyst, calcified fibroadenoma, fat-containing lesion, or other finding the radiologist wants to document so that future readers of your imaging know it is there and has already been evaluated. Like Category 1, the recommendation is routine screening. The distinction between 1 and 2 matters mainly for record-keeping: Category 2 tells the next radiologist “I saw this, and it’s nothing to worry about.”

Category 3 and the Probably Benign Finding

Category 3 is where things get more nuanced. A BI-RADS 3 assessment means the radiologist has identified a finding that is almost certainly benign but not definitively so. By definition, the expected likelihood of malignancy is no more than about 2%. Rather than jumping straight to biopsy, the standard recommendation is short-interval follow-up imaging, typically at six months, then again at 12 and 24 months, to confirm the finding stays stable.

This wait-and-watch approach makes sense when you consider the math. A study of over 1,100 patients with BI-RADS 3 lesions found on screening breast ultrasound reported an overall malignancy rate of 0.7%.⁴ That is low enough that biopsying every Category 3 finding would subject a large number of people to an invasive procedure for no benefit. If follow-up imaging shows the finding is unchanged after two years, it typically gets downgraded to Category 2.

There is a significant caveat, though. The 2% threshold does not hold equally well for all age groups. A large analysis using the National Mammography Database found that cancer yield among BI-RADS 3 mammographic findings rose steadily with age, exceeding the 2% benchmark in women 60 and older and reaching about 4.6% in women in their 80s.⁵ This has led some experts to question whether Category 3 should be used more cautiously in older patients, where biopsy rather than surveillance might be more appropriate. If you are over 60 and receive a Category 3 result, it is worth discussing with your physician whether closer follow-up or biopsy makes more sense given your individual risk factors.

Category 4 and Its Three Subdivisions

Category 4 covers findings that are suspicious enough to warrant a biopsy but span a wide range of concern. The probability of malignancy stretches from just above 2% all the way up to just under 95%, which is an enormous gap. To make this category more useful, the system breaks it into three subdivisions.⁶

4A (low suspicion): Greater than 2% but no more than 10% likelihood of malignancy. Findings here often turn out to be benign on biopsy. A palpable, partially circumscribed solid mass or a complicated cyst might land in 4A.
4B (moderate suspicion): Greater than 10% up to 50% likelihood. The finding has some worrisome features but is not a slam-dunk concern. Biopsy is clearly warranted, and results could go either way.
4C (high suspicion): Greater than 50% but less than 95% likelihood. Findings here look worrisome, though they do not quite meet the threshold for Category 5. Most radiologists expect these to come back malignant on biopsy, but there is enough uncertainty that they stop short of the highest category.

These subdivisions matter for clinical decision-making. One study of breast MRI lesions classified as Category 4 found that the actual positive predictive values for the three subdivisions were roughly 2%, 12%, and 68% for 4A, 4B, and 4C respectively.⁷ Knowing which subdivision you fall into can help set expectations about what the biopsy is likely to find and can influence how quickly the procedure is scheduled. All three subdivisions carry a recommendation for tissue sampling, but the emotional weight of a 4A is quite different from a 4C.

Categories 5 and 6

Category 5 is reserved for findings that are almost certainly malignant, with a positive predictive value of 95% or higher. These lesions typically show multiple suspicious features on imaging. One study suggested that a Category 5 assessment is justified when at least four suspicious descriptors are identified across diagnostic mammography and ultrasound.⁸ A spiculated mass with associated suspicious calcifications and skin thickening, for instance, would typically earn a Category 5. Biopsy is essential, and treatment planning often begins in parallel because the probability of a cancer diagnosis is very high.

Category 6 is straightforward and sometimes confuses people who encounter it for the first time. It is used when a malignancy has already been proven by biopsy and additional imaging is being performed for other reasons, such as evaluating the extent of disease before surgery, checking response to chemotherapy, or screening the opposite breast. Category 6 does not represent a new finding; it is a label that keeps the known cancer documented in the reporting system so that the imaging is not mistakenly interpreted as a new suspicious finding.

How Breast Density Fits In

Your BI-RADS report will typically include a breast density classification alongside the assessment category. Breast density is scored on a four-letter scale (a through d), ranging from almost entirely fatty tissue to extremely dense tissue. This matters because dense breast tissue can mask cancers on a mammogram, making them harder to detect.

Research from a large UK cohort showed that mammographic sensitivity dropped from about 75% in the least-dense breasts (category a) to roughly 51% in the densest breasts (category d). Interval cancer rates, meaning cancers found between screening rounds, were also significantly higher in dense breasts, reaching about 8 per thousand screens in the densest category compared to fewer than 2 per thousand in the least dense.⁹ Separate research in a digital mammography screening program confirmed that sensitivity drops substantially in the densest breast category compared to all others.¹⁰

In many parts of the United States, laws now require that patients be notified if they have dense breast tissue. If you are told you have heterogeneously or extremely dense breasts (categories c or d), this does not mean anything is wrong. It means that mammography alone may miss some findings in your tissue, and you and your doctor may want to discuss supplemental screening with ultrasound or MRI. Dense breast tissue is also an independent risk factor for breast cancer, separate from its masking effect.

BI-RADS Across Different Imaging Types

BI-RADS originally applied only to mammography, but the system has expanded. The fourth edition, released in 2003, added separate lexicons for breast ultrasound and breast MRI for the first time.¹¹ The fifth edition, the most recent, significantly expanded the MRI section and added new ultrasound terms to reflect advances in technology like elastography.¹²

The numerical categories (0 through 6) mean the same thing across all three modalities: a Category 4 on ultrasound carries the same management recommendation as a Category 4 on mammography or MRI. However, the specific descriptors radiologists use differ by modality because each imaging type shows different tissue characteristics. MRI, for example, evaluates enhancement patterns after contrast injection, while ultrasound assesses features like echogenicity and margins. The assessment categories translate across modalities, but the vocabulary for arriving at those assessments is modality-specific.¹³

How Much Do Radiologists Agree on Scores?

One question patients rarely think to ask is how consistent these scores are between different radiologists reviewing the same images. The answer depends on the type of finding. A study examining interobserver agreement on breast MRI found substantial overall agreement on BI-RADS category assignment, but agreement was notably weaker for certain types of findings. Radiologists agreed on the morphology of enhancement in about 71% of cases, and overall category agreement was also substantial. However, when the finding was a non-mass enhancement (a pattern that does not form a discrete lump), agreement on the BI-RADS category dropped to only moderate levels.¹⁴

This variability is worth knowing about. If you receive a borderline assessment, particularly one that sits at the boundary between Category 3 and Category 4, the assessment could reasonably differ between two qualified radiologists. Seeking a second opinion from a breast imaging specialist is not unreasonable in ambiguous cases, especially when the decision is between continued surveillance and biopsy.

When BI-RADS Does Not Apply Well

BI-RADS was designed around the adult female breast, and applying it outside that population can lead to problems. A study examining the use of BI-RADS ultrasound classification in pediatric and adolescent patients concluded that the system overestimates the likelihood of malignancy in this group. Breast masses in children and teenagers are overwhelmingly benign, and the study recommended that BI-RADS scores should not be assigned to pediatric patients and that BI-RADS-defined biopsy recommendations should be disregarded in that population.¹⁵ If your teenager has a breast lump and receives imaging, ask whether the assessment was made using pediatric-specific criteria rather than the standard BI-RADS framework.

Male breast imaging presents a similar mismatch. While BI-RADS categories are sometimes applied to male breast evaluations, gynecomastia (benign breast tissue enlargement) is by far the most common finding in men, and the descriptors and risk thresholds in BI-RADS were calibrated on female breast imaging data. Radiologists evaluating male breast findings typically rely on clinical context more heavily than the standardized BI-RADS malignancy thresholds.

Misdiagnosis and Medicolegal Considerations

Errors in breast imaging are a significant source of malpractice claims. According to a statement from the European Society of Breast Imaging, detection errors or misapplications of the BI-RADS lexicon or score account for about 5% of malpractice claims in the field.¹⁶ The most common problems include radiologist distraction or fatigue, satisfaction of search (stopping the review after finding one abnormality and missing a second), and failure to follow the systematic approach the lexicon was designed to enforce.

For patients, the practical takeaway is that BI-RADS is a tool, not a guarantee. A Category 1 or 2 assessment is reassuring and correct the vast majority of the time, but no imaging system catches every cancer. If you develop a new symptom, such as a palpable lump, skin change, or nipple discharge, between screening rounds, do not wait for your next scheduled mammogram just because your last result was normal. Interval symptoms warrant a diagnostic evaluation regardless of your most recent BI-RADS score.

Artificial Intelligence and BI-RADS

AI-assisted mammography reading is increasingly being tested alongside human radiologists, and much of the research frames AI performance in BI-RADS terms. One system evaluated against a large dataset appropriately reclassified about 83% of false-positive cases (those originally scored as BI-RADS 4B or 4C but ultimately found to be benign) into a benign group, and it correctly flagged about 54% of false-negative cases (originally scored as BI-RADS 3 or 4A but later confirmed as malignant) as needing further attention.¹⁷ These numbers suggest AI can help reduce both unnecessary biopsies and missed cancers, though neither problem is eliminated entirely.

An emerging concern is how AI results affect patients when they appear in health portals. Research published in a digital medicine journal found that including AI scores alongside BI-RADS results in patient-facing reports significantly increased anxiety, even when the AI agreed with the radiologist’s assessment. Patients who saw an AI score close to a threshold for concern reported higher worry about breast cancer, even after a normal BI-RADS 1 determination.¹⁸ How to present AI results to patients without triggering unnecessary alarm is an active area of discussion in radiology.

Breast Density Guidelines Around the World

While BI-RADS itself is used internationally, the guidelines for what to do about dense breast tissue vary widely by country. A systematic review of breast screening guidelines worldwide found that the overwhelming majority of clinical guidelines addressing mammographic density came from high-income countries. Only a handful came from middle-income countries, and low-income countries with significant breast cancer burdens had virtually no tailored guidelines for managing dense breasts in screening.¹⁹ This disparity means that a woman with dense breasts in a well-resourced country may be offered supplemental MRI screening, while a woman with the same tissue composition in a lower-income setting may not even receive notification about her density classification. The BI-RADS framework provides the language to describe and categorize density globally, but the clinical response to that information remains uneven.