How do I make a Bland–Altman plot from paired measurements?

Paste a table with two numeric columns (method 1 / reference and method 2 / new), or upload CSV/TSV. Keep Plot type on Difference. The figure is difference versus mean with bias and 95% limits of agreement. Download the SVG for a report or slide.

What are 95% limits of agreement?

They are the mean difference plus and minus 1.96 times the sample SD of the paired differences. If differences are approximately normal and independent, about 95% of them fall in that band. This tool uses that textbook construction. It does not compute mixed-effects or heteroscedastic LoA.

Is a Bland–Altman plot the same as an MA plot?

No. An MA plot is RNA-seq / microarray log-ratio versus mean expression. A Bland–Altman plot is clinical method agreement. They share a mean-versus-difference idea and nothing else. Use the MA plot generator for DESeq2 tables.

Does this replace MedCalc or Passing–Bablok?

No. It only plots the pairs you paste. It does not run Passing–Bablok, Deming regression, or repeated-measures ANOVA. For publication method-comparison, use software that matches your study design and report confidence intervals on the LoA.

What if the cloud fans out or slopes with the mean?

That is proportional or heteroscedastic disagreement. This page still draws classic horizontal LoA and prints the OLS slope of difference versus mean as a diagnostic. It does not switch to regression-based limits. Decide clinically whether a percent-difference plot is more honest for your scale.

Is the Bland–Altman plot generator free?

The precise, table-driven plot is free to use and free to download as SVG, with no sign-up. Sign in only if you want AI illustrations or other ConceptViz generation tools.

Research Tool

Bland-Altman Plot Generator for method comparison

Free Bland-Altman plot generator. Paste paired measurements to draw bias and 95% limits of agreement, then download SVG.

Paste paired method A and BDifference or percent differenceBias and 95% limits of agreementDownload a watermark-free SVG — free

Paste paired measurements — renders an exact mean-difference plot as SVG, free

Paired measurements

Paste two columns (method A / reference and method B / new). The tool does not run method-comparison software.

40 pairs parsed

Plot typeTitle

n=40 · bias=-1.75 · SD=3.63 · LoA -8.87 to 5.37 · difference vs mean slope=-0.009 (R²=0.003). Limits of agreement are mean ± 1.96 SD of the paired differences, not a mixed-effects or Passing–Bablok analysis.

Bland–Altman Plot Examples

The first four are exact engine renders from tables. The last two are AI illustrations of Illustration mode.

View:

Cuff vs monitor

Exact engine render — difference versus mean, bias ± 1.96 SD.

bland-altmanblood-pressureloa

Lab vs glucometer

Exact engine render — glucometer reads high, so A − B bias is negative.

glucosebias

Percent difference

Exact engine render — y = 100 × (A − B) / mean, still classic horizontal LoA.

percentloa

Proportional bias

Exact engine render — slope of difference vs mean is a diagnostic; LoA stay horizontal.

proportionalbias

AI illustration — agreement

AI illustration of Illustration mode — visual mockup, not plotted from pairs.

aibland-altmanillustration

AI illustration — LoA schematic

AI illustration of Illustration mode — visual mockup, not a fitted LoA.

ailimits-of-agreementillustration

What is a Bland–Altman plot?

A Bland–Altman plot (also called a Tukey mean-difference plot) compares two methods that measure the same continuous quantity. For each pair it plots the difference (A − B) against the mean ((A + B) / 2), then draws the mean difference (bias) and the 95% limits of agreement at bias ± 1.96 SD. Correlation can look excellent while the methods still disagree by a clinically important amount — that is the point of this figure. This page draws that plot from the table you paste. It does not run MedCalc, Passing–Bablok, Deming regression, or mixed-effects limits for repeated measures, and it is not an RNA-seq MA plot.

How bias and limits of agreement are read

Paste two paired columns. Method A is usually the reference (lab, cuff, gold standard); method B is the new device. Difference is A − B, so a negative bias means B reads higher.
The solid line is the mean difference. The dashed lines are mean ± 1.96 times the sample standard deviation of the differences (n − 1). About 95% of future differences are expected to fall between them if the differences are roughly normal and independent.
Whether those limits are acceptable is a clinical decision you make before looking at the plot, not a p-value this page computes.
Percent difference mode uses 100 × (A − B) / mean. Rows whose mean is zero cannot be converted and are skipped. Classic horizontal LoA are still used; a large slope of difference versus mean is printed as a diagnostic, not as regression-based limits.

Bland–Altman versus an MA plot

Both plot a difference against a mean, but they are different figures. Bland–Altman is clinical method comparison with limits of agreement. An MA plot is the genomics figure of log-ratio versus average intensity for thousands of genes. Use this page for paired assays; use the MA plot generator for DESeq2 / limma results.

Export

Download a scalable SVG with no watermark and no account required for the precise plot. The AI Illustration tab is only for slides when you do not have pairs yet — it will not match your bias or LoA.