- 01One number is a snapshot with built-in noise
A lab value carries your body's real day-to-day swings plus the lab's own measurement error, so a small gap between two tests is often just that wobble rather than a true change. The direction across several readings is what carries the signal.
- 02A change has to clear the noise before it counts
Each marker has its own natural amount of wobble. Triglycerides can swing about 56% between two honest draws and still be unchanged; total cholesterol only about 18%. A move smaller than that is likely noise, not news.
- 03That is why one odd result is rechecked, not treated
Extreme first readings tend to drift back toward your own average on a repeat, an effect called regression to the mean, so guidelines confirm an abnormal result before acting on it.
Your blood test numbers move between draws mostly because of normal biological ups and downs, the lab's own measurement error, and everyday things like whether you fasted, how hydrated you were, and even standing versus lying down while your blood was taken. In most cases your health did not flip overnight. That is why the direction across several readings tells you far more than any single number, and why a clinician usually rechecks one borderline result before acting on it.
Why do my blood test results change between tests?
Three separate sources of change stack on top of each other every time blood is drawn. None of them means your health shifted.
- Biological variation: your body's own day-to-day and hour-to-hour swings. Hormones, meals, sleep, exercise, and the time of day all nudge the number. This within-person variation has a name in lab medicine, CVi, and it has been measured for hundreds of markers.1
- Analytical imprecision: the lab's own measurement error. Run the exact same tube twice and the instrument returns two slightly different numbers. Good labs keep this small, but it is never zero.
- Pre-analytical factors: everything that happens before the lab measures anything. Fasting or not, how hydrated you are, how long the tourniquet was on, and posture. Going from lying down to standing concentrates your blood slightly (hemoconcentration) and can raise protein-bound markers and cell counts by roughly 5 to 10%.6
How much can blood tests vary?
It depends entirely on the marker, and the spread is enormous. Your body defends some values to a fraction of a percent and lets others roam. The table below shows typical within-person variation (CVi) and how big a change has to be before it clears the combined biological and lab noise. That change threshold has a name, the reference change value (RCV), and it works out to roughly 2.77 times the combined variation.2
| Marker | Typical day-to-day variation (CVi) | How big a change must be to count as real (RCV) |
|---|---|---|
| Sodium | ~0.6% | ~2 to 3% |
| HbA1c (3-month blood sugar) | ~1.5 to 2% | ~5% |
| Fasting glucose | ~5 to 6% | ~15 to 17% |
| Total cholesterol | ~6% | ~18% |
| TSH (main thyroid test) | ~19 to 20% | ~55% |
| Triglycerides (a blood fat) | ~20% | ~56% |
The variation figures come from the EFLM Biological Variation Database;1 the change thresholds use the reference change value formula.2 Both are approximate and shift a little from lab to lab, but the ranking holds: a 15% jump in TSH or triglycerides is unremarkable, while the same 15% jump in sodium would be a genuine signal.
What one reading tells you versus what the trend tells you
Is one abnormal blood test a problem?
Usually it is a prompt to look again, not a diagnosis. Two things make a lone out-of-range value weak evidence on its own. First, broad panels flag by chance: run enough tests and one will land outside its range even in a perfectly healthy person. Second, extreme readings tend to move back toward your own average when you repeat them, an effect called regression to the mean.5 An unusually high first result is, on average, partly a bad-day fluke that a second test walks back.
An out-of-range flag is a prompt to look closer, not a diagnosis.
That is the same reading habit how to read any lab result walks through in full. It also connects to why the in-range line is blurry at the edges rather than a hard wall: as reference ranges explained puts it, about 1 in 20 healthy people fall outside a given range on any single test.
Should I repeat a blood test, and why would a doctor repeat one?
Because a confirmed change is worth acting on and a one-off blip is not. Major guidelines build the repeat straight into the rules. For diagnosing diabetes, the American Diabetes Association says that without clear symptoms of very high blood sugar, an abnormal result should be confirmed by a second test before the diagnosis stands.4 The repeat is how you separate a real change from the noise described above.
Some of that noise is avoidable, and knowing which rules apply keeps a repeat honest. Lipid panels are a good example. For years a cholesterol test meant fasting first, but that turns out to change the numbers little for most people. The 2016 European Atherosclerosis Society and EFLM joint consensus put it plainly:3
Fasting is not routinely required for determination of a lipid profile.Nordestgaard BG et al., EAS/EFLM joint consensus statement, European Heart Journal, 2016
The practical point: when you do repeat a marker, keep the conditions comparable. Same rough time of day, same fasting state if it matters for that test, same lab where you can. Comparing a fasting result with a non-fasting one, or a morning draw with an evening one, adds noise you could have avoided.
What a meaningful change looks like
Put the pieces together and a simple rule falls out. Before you treat a difference between two results as real, check whether it beat that marker's reference change value. A total cholesterol that drops from 210 to 195 mg/dL (about 7%) has not cleared the roughly 18% bar, so on its own it could be noise. A triglyceride reading that falls from 200 to 150 (25%) still has not cleared its roughly 56% bar. These figures are worked illustrations of the general rule, not a way to interpret your own results on your own; bring your actual numbers to your clinician. More readings cut through the noise better than any single big number can: three or four comparable draws where the direction holds.
How often to retest depends on the marker and why it is being watched, and that is a conversation for your clinician rather than a fixed rule. As a rough sense of scale, cholesterol and HbA1c are typically rechecked over months, not days, because change that slow needs spacing to show up above the wobble. Repeating a stable marker every week mostly measures the noise.
One reading tells you where to look. Your last few readings, and the direction they point, tell you whether anything has changed. When you get a new result, ask for the previous ones and read them together.
Link · European Federation of Clinical Chemistry and Laboratory MedicineEFLM Biological Variation DatabaseLook up the typical day-to-day variation and reference change value for hundreds of individual markers.biologicalvariation.eu
Citations
- EFLM Biological Variation Database, European Federation of Clinical Chemistry and Laboratory Medicine.
- Fraser CG, Harris EK. Generation and application of data on biological variation in clinical chemistry. Critical Reviews in Clinical Laboratory Sciences. 1989;27(5):409-437.
- Nordestgaard BG, Langsted A, Mora S, et al. Fasting is not routinely required for determination of a lipid profile: EAS/EFLM joint consensus statement. European Heart Journal. 2016;37(25):1944-1958.
- American Diabetes Association. Diagnosis and Classification of Diabetes: Standards of Care in Diabetes 2024. Diabetes Care. 2024;47(Suppl 1):S20-S42.
- Bland JM, Altman DG. Statistics Notes: Regression towards the mean. BMJ. 1994;308(6942):1499.
- Rifai N, Horvath AR, Wittwer CT, eds. Tietz Textbook of Clinical Chemistry and Molecular Diagnostics. 6th ed. Elsevier; 2018. Preanalytical variation (posture, hemoconcentration).
Educational context only, not medical advice or a diagnosis. Always discuss your results with a clinician.