Cancer screening blood tests detect late-stage cancer well and early-stage cancer poorly. In the guides we publish here, that inversion is the fact readers are most surprised by, because the entire category is sold on early detection.
The reason is physical. These tests look for tumour DNA or tumour cells circulating in blood, and a small early tumour sheds very little of either. Detection rates climb steeply with stage, which means the tests are weakest exactly where finding a cancer would change the most.
The Verdict
The Three Numbers That Decide Accuracy
Four measures describe a screening test, and manufacturers reliably quote the two that flatter and omit the one you care about.
| Measure | What it answers | What it means in practice | Why it misleads |
|---|---|---|---|
| Sensitivity | Of people who have cancer, the share the test flags | A test with 52 percent sensitivity misses roughly half the cancers present at the moment it is run | Almost always quoted as a single figure averaged across all stages, which hides how poor it is at stage I |
| Specificity | Of people who do not have cancer, the share the test correctly clears | At 99.5 percent specificity, 1 in 200 healthy people still gets a positive result | Sounds close to perfect and is the number that generates most false alarms, because most people screened are healthy |
| Positive predictive value | Of people who get a positive result, the share who actually have cancer | This is the number you personally care about, and it is the one least often quoted | It depends on how common cancer is in the group being screened, so it changes with age and risk profile |
| Negative predictive value | Of people who get a negative result, the share who genuinely have no cancer | High for every test in this category, because most people screened do not have cancer | Reassuringly high for a misleading reason. A negative result mainly reflects how rare cancer is rather than how hard the test looked |
Positive predictive value is the number that answers your actual question, which is what a positive result means for you specifically. It is also the only one of the four that changes with who is being screened, and that dependency is why no manufacturer can print a single figure for it on a box.
Sensitivity Falls Sharply at Early Stages
A headline sensitivity figure is an average across every stage of every cancer the test covers. The underlying distribution is uneven enough that the average describes almost nobody.
Detection at stage IV, where tumour burden is high and shedding into the bloodstream is substantial, runs far above the average. Detection at stage I, where a tumour may be a few millimetres across, runs far below it. A test quoted at roughly half of all cancers detected will find a much smaller share of the stage I cancers it encounters.
Cancer type matters too. Aggressive cancers that shed heavily, including pancreatic and ovarian, are detected more readily than slower, less vascular tumours. That is a mixed blessing: the cancers these tests find best are also the ones where earlier detection has the strongest case, and the ones they find worst include several already covered by existing screening programmes.
What Each Test's Numbers Look Like
Three tests dominate the questions we get, and they are doing meaningfully different jobs.
| Test | What it screens for | Reported performance | The gap in the numbers | Regulatory status |
|---|---|---|---|---|
| Galleri (GRAIL) | Signal from more than 50 cancer types | About 52 percent sensitivity across all stages, with 99.5 percent specificity | Stage-dependent performance. Detection at stage I is far below the headline figure | Prescription test, self-pay at around $949. An FDA advisory committee review is scheduled for 23 September 2026 |
| Shield (Guardant Health) | Colorectal cancer only | 83 percent sensitivity for colorectal cancer, 90 percent specificity for advanced neoplasia | About 13 percent sensitivity for advanced precancerous lesions, which is the part colonoscopy is for | FDA-approved in July 2024 as a primary colorectal screening option for average-risk adults 45 and older |
| Trucheck (Datar Cancer Genetics) | Circulating tumour cell clusters across multiple cancer types | Company-published performance figures, without an FDA approval or a US registrational trial of the scale behind the other two | Independent validation in a US screening population is missing | Sold direct to consumers in several markets. Treat published figures as manufacturer data |
Guardant Health's Shield test was approved by the US Food and Drug Administration in July 2024 on the basis of the ECLIPSE registrational study in more than 20,000 average-risk adults. That is the strongest evidence base in the category, and it covers one cancer.
GRAIL's Galleri covers signal from more than 50 cancer types and has the largest published multi-cancer dataset. Our Galleri review works through its pricing and its detection claims in detail, and our news brief on the scheduled FDA advisory committee review covers what that meeting can and cannot change for a buyer.
Why a Positive Result Is Often Wrong
Positive predictive value is set by how common cancer is in the group being screened, and in general adult screening populations cancer is rare. That rarity turns a small false positive rate into a large share of the positives.
Two things raise it. Screening older people raises it because cancer incidence rises with age. Screening people with a family history or a known risk factor raises it for the same reason. This is why the same test performs better in a 68-year-old with a strong family history than in a healthy 42-year-old, without anything about the assay changing.
It also explains why these tests are prescription products rather than direct-to-consumer kits in most markets. The decision about whether the arithmetic works for you depends on information a clinician has and a website does not.
The Cost of a False Positive
A false positive is not a neutral event that resolves with a phone call. It initiates a diagnostic work-up, and multi-cancer tests direct that work-up using a prediction of where the signal came from.
- Imaging, often more than one modality. Cost, radiation exposure in some cases, and incidental findings that generate their own follow-up.
- Endoscopy or biopsy where imaging is inconclusive. Procedural risk, sedation, and recovery time.
- Weeks of waiting. Resolution of a false positive frequently takes longer than confirmation of a true one, because ruling something out requires looking in more places.
- Cost that the test price did not include. The work-up is billed separately, and insurance coverage of it varies.
None of this argues against screening. It argues for going in having decided in advance that you are willing to complete the work-up if the result comes back positive, because a positive result you then decline to investigate is the worst of both outcomes.
Who These Tests Are Accurate Enough For
Three groups get the most out of the current generation, and two get the least.
- Older adults with a family history. Higher background incidence lifts positive predictive value, which is the number that determines whether a positive is worth acting on.
- People at risk of cancers with no screening programme. Pancreatic and ovarian cancer have no population screening test, and this is where a multi-cancer approach adds something rather than duplicating.
- People who will not have a colonoscopy at all. For colorectal screening specifically, Shield's comparison is against no screening rather than against colonoscopy, and it wins that comparison.
Two groups should think harder. Anyone under 45 with no family history is screening a population where cancer is rare enough that most positives will be false. Anyone who would not proceed with imaging and biopsy after a positive result should not take the test, since the result would change nothing except their anxiety.
Our reading would change if a randomised trial showed a multi-cancer blood test reducing cancer-specific mortality. Every figure discussed here describes detection rather than outcome, and detecting a cancer earlier only helps if treating it earlier changes what happens. The NHS-Galleri trial in the United Kingdom was designed to test exactly that, and its mortality readout is the evidence this category still needs.
Before ordering one of these tests, ask your clinician what your positive predictive value would be given your age and family history, since that single number decides whether the result will be worth acting on.
Frequently Asked Questions
How accurate are cancer screening blood tests?
Accuracy depends on which number you mean and which cancer you are asking about. Galleri reports about 52 percent sensitivity across all stages with 99.5 percent specificity, which means it misses roughly half of cancers present and produces a false alarm in about 1 in 200 healthy people screened. Guardant's Shield test reports 83 percent sensitivity for colorectal cancer specifically, and about 13 percent for the advanced precancerous lesions colonoscopy is designed to remove.
Can a blood test detect cancer early?
It can, and it does so far less reliably than the marketing implies. Detection depends on tumour DNA or cells circulating in blood at levels the assay can find, and small early-stage tumours shed very little. Sensitivity in these tests rises steeply with stage: detection at stage IV is far higher than at stage I. That is the inverse of what an early detection test needs to do, and it is the central limitation of the whole category rather than a flaw in any one product.
What does a positive cancer blood test mean?
It means a diagnostic work-up follows. It does not mean you have cancer. Positive predictive value in this category runs well below 100 percent, and in a general screening population a meaningful share of positives turn out to be false alarms. The next step is imaging and often a biopsy directed by the test's prediction of where the signal came from. Nothing is confirmed until tissue or imaging confirms it, and a positive result on its own is not a diagnosis.
Why is 99.5 percent specificity not good enough?
Because almost everyone screened is healthy, so even a small false positive rate produces a large number of false alarms in absolute terms. Screen 10,000 people with a test that is 99.5 percent specific and roughly 50 healthy people get a positive result. If cancer is present in 50 of those 10,000 and the test catches half of them, you have 25 true positives against 50 false ones. The arithmetic decides this, and how rare cancer is sets it.
Do cancer screening blood tests replace colonoscopy or mammography?
No, and no manufacturer claims they do. Shield is FDA-approved as a primary colorectal screening option, and its low sensitivity for advanced precancerous lesions means it cannot do what colonoscopy does, which is find and remove polyps before they become cancer. Multi-cancer tests are positioned as additive to existing guideline screening rather than as a substitute. Dropping a mammogram or a colonoscopy because a blood test came back clear removes a screen that works.
What happens if the blood test is wrong?
A false positive leads to a diagnostic work-up that finds nothing: imaging, sometimes endoscopy, sometimes a biopsy, and the weeks of waiting that go with them. That carries financial cost, procedural risk and real anxiety. A false negative is the more dangerous error, because it can produce false reassurance that delays acting on a symptom. Neither is rare enough to treat as theoretical, which is why the decision to screen belongs with a clinician who knows your risk profile.
Is Shield as good as a colonoscopy?
For finding cancer that is already present, Shield performs within range of other non-invasive screening options at 83 percent sensitivity. For preventing cancer, it does not compare: colonoscopy finds and removes advanced precancerous lesions, and Shield detects only about 13 percent of them. The two tests are doing different jobs. Shield's strongest case is for people who will not have a colonoscopy at all, where the comparison is against no screening rather than against the gold standard.
Which cancer screening blood test is most accurate?
For colorectal cancer specifically, Shield has the strongest evidence, with an FDA approval based on a registrational trial of more than 20,000 people. For multi-cancer screening, Galleri has the largest published dataset and is the only one heading to an FDA advisory committee. Tests without an FDA submission or independent validation in a screening population should be read as manufacturer-reported performance until someone outside the company confirms it.
Related
- Galleri review: the multi-cancer test in detail
- Best multi-cancer blood test: the options compared
- Colon cancer blood test vs colonoscopy: what each one finds
- Galleri at the FDA advisory panel: what the review changes
- At-home blood test accuracy: the collection-method question
- Full body MRI: the imaging side of whole-body screening
- Prenuvo review: the imaging alternative reviewed