Figure Legends and Captions: What Must Be In Them

A figure legend (also called a figure caption) is the text that travels with a figure so a reader can understand it without going back to the Methods or Results section. Reviewers and editors check legends against a fairly consistent checklist across journals: a title stating the main finding, a description of every panel, a definition of every abbreviation and symbol on the image, the statistical test used, what error bars represent, and the sample size. A figure that is otherwise publication-ready is routinely sent back at technical check, or flagged by a reviewer, because its legend is missing one of these elements — not because the figure itself is wrong.

This guide gives a complete, annotated example of a figure legend, a checklist of the elements journals expect, and the specific wording conventions (n=, error-bar definitions, statistical-test phrasing) that make a legend self-contained.

Worked example: a complete figure legend

Here is a legend for a two-panel bar-chart figure, annotated element by element. This is the level of detail a biomedical or life-sciences journal expects; the same structure applies with lighter statistical detail to figures in other fields.

Figure 2. Drug treatment reduces cell viability in a dose- and time-dependent manner. (A) Viability of HeLa cells 24 hours after treatment with compound X at five concentrations (0, 1, 5, 10, and 25 µM; n = 6 independent experiments per condition). (B) Viability of the same cell line at 48 hours under identical treatment conditions (n = 6). Viability was measured by MTT assay and is expressed as a percentage of the untreated control. Error bars represent standard deviation (SD). Statistical significance was assessed by one-way ANOVA with Tukey’s post hoc test comparing each treatment group to control (*p < 0.05, **p < 0.01, ***p < 0.001). Scale bar in inset micrographs = 50 µm.

Reading only that paragraph, without the manuscript text, a reader can determine: what was measured, in which cell line, at how many concentrations and time points, with how many replicates, using which assay, what the error bars mean, which statistical test produced the significance markers, and the scale of the inset images. That is the standalone-comprehension bar every complete legend has to clear.

The elements a complete figure legend needs

Element	What it does	Example wording
Title sentence	A short declarative sentence stating the figure’s main point — not just a topic label	“Drug treatment reduces cell viability in a dose- and time-dependent manner.”
Panel descriptions	What each lettered panel shows, keyed to the panel labels used on the image itself	“(A) Viability at 24 hours… (B) Viability at 48 hours…”
Sample size (n)	The number of independent replicates, animals, participants, or specimens behind each data point or bar	“n = 6 independent experiments per condition”
Method/assay used	The technique that produced the data shown, if not already unambiguous from context	“Viability was measured by MTT assay”
Units and normalization	What the axis or value actually represents	“expressed as a percentage of the untreated control”
Error bar definition	What the error bars or shaded region represent — SD, SEM, and 95% CI are not interchangeable and journals require the specific one used	“Error bars represent standard deviation (SD)”
Statistical test	The exact test and correction/post hoc method applied, plus what the significance markers mean	“one-way ANOVA with Tukey’s post hoc test… *p < 0.05”
Abbreviation/symbol key	Every abbreviation, color code, and line style used on the figure, defined at first use in the legend even if already defined in the main text	“DAPI (4′,6-diamidino-2-phenylindole, nuclear stain)”
Scale bar value	The real-world distance a scale bar on a micrograph or image represents	“Scale bar = 50 µm”

Not every figure needs every element — a conceptual diagram with no data points does not need a statistical-test line, and a single-panel figure does not need panel labels. The test is whether a reader who has not read the Methods section can still interpret the figure correctly using only the legend.

Figure legend vs. figure caption vs. in-text callout

These three terms get used loosely and interchangeably, but they describe different things:

• Figure legend is the term most common in biomedical, life-sciences, and STM journal style (and in ICMJE- and CSE-aligned author guidelines) for the full explanatory block of text accompanying a figure.
• Figure caption is used interchangeably with “legend” by many journals and style guides — APA Style, for instance, uses “caption” for the same element and requires it to appear directly below the figure, with a separate bolded figure number and italicized brief title line above the explanatory text. There is no universal rule distinguishing the two terms; check the specific journal’s or style guide’s author instructions rather than assuming one implies a shorter or less complete version of the other.
• In-text callout or reference (e.g., “as shown in Figure 2,” “see Fig. 3B”) is different from either — it is a pointer embedded in the manuscript’s running prose that directs the reader to look at the figure. It is not a substitute for the legend and carries none of the standalone explanatory content the legend is responsible for.

Where legends go: placement conventions vary by journal

Placement is one of the more common technical-check failures because it is journal-specific and easy to get wrong when reusing a template from a different publisher:

• Legend directly beneath the figure — the convention for most journal websites and final typeset PDFs, and for APA-style manuscripts throughout.
• All legends collected on a separate “Figure Legends” page, placed after the references or tables, with the figures themselves submitted as separate image files — a common requirement at initial submission for many biomedical journals, so that figures can be reviewed at full size without legend text overlapping the image.
• Legends embedded in a combined figures-and-tables file appended to the manuscript, used by some journals specifically for the peer-review version, then split apart for production if the paper is accepted.

Always check the specific journal’s current author guidelines before submission — this is one of the details that a technically correct legend can still fail on if it is placed according to the wrong convention. See CASRAI’s guide to scientific figure rules for journal submission for the DPI, file-format, and panel-labeling requirements that apply alongside legend placement.

A figure legend checklist before submission

• Does the first sentence state the figure’s main point, not just its topic (“Cell viability decreased with drug concentration,” not “Bar chart of cell viability”)?
• Is every lettered panel described, in order, keyed to the panel labels actually printed on the figure?
• Is the sample size (n) stated for every panel that shows aggregated or averaged data?
• Is every abbreviation, symbol, and color/line code on the figure defined in the legend, even if already defined earlier in the manuscript text?
• Is it stated explicitly whether error bars represent SD, SEM, or a confidence interval — never left implicit?
• Is the statistical test named, along with any multiple-comparisons correction or post hoc method, and what the significance markers (e.g., *, **, ***) mean?
• Is a scale bar’s real-world value given, for any micrograph, gel, or blot image?
• Does the legend avoid introducing new results or interpretation not discussed in the Results text — a legend explains what is shown, it does not argue a conclusion?
• Is the legend placed according to the target journal’s current author guidelines (below the figure vs. a separate legends page)?

Common mistakes

Title-only captions. “Figure 4. Western blot results.” with no panel description, no n, and no explanation of what the bands represent fails the standalone-comprehension test — a reader cannot interpret the figure from the legend alone, which is exactly what a legend exists to make possible.

Undefined error bars. Reporting bars without stating whether they represent SD, SEM, or a confidence interval is one of the most common reviewer comments on figure legends, because the three represent meaningfully different things about the same data and cannot be inferred from the chart alone.

Reusing abbreviation definitions from the main text only. A figure is often viewed independently of the surrounding manuscript — in a preprint aggregator, a slide deck, or a reader who jumps straight to the results — so abbreviations need to be defined in the legend itself, not only on their first appearance in body text.

Legends that argue a conclusion. A legend describes what the figure shows (methods, n, statistics); the interpretation of what that means belongs in the Results or Discussion text, not the legend.

Frequently asked questions

What is the difference between a figure legend and a figure caption?

In most journal and biomedical-style usage, the two terms are interchangeable — both refer to the explanatory text accompanying a figure. Some style guides (APA among them) formally use “caption,” while ICMJE- and CSE-aligned biomedical journal style tends to use “legend.” Check the specific journal’s or style guide’s instructions rather than assuming a distinction; neither term implies a shorter or less complete version of the other.

How long should a figure legend be?

There is no universal word limit — length is driven by how many elements a specific figure requires (panel count, statistical detail, number of abbreviations), not by a fixed target. A single-panel schematic may need only a sentence or two; a six-panel data figure with several statistical comparisons may reasonably run to a full paragraph. Some journals do cap total legend length or specify a maximum per figure in their author guidelines; check before finalizing a long legend.

Do figure legends need to state the sample size?

Yes, for any panel showing averaged, aggregated, or statistically compared data. Reviewers and readers need the sample size to judge whether a result is adequately powered and to interpret the error bars and statistical test correctly — a legend without n is a common reviewer request for revision.

What goes in a legend for a multi-panel figure?

An overall title sentence describing the figure’s main point, followed by a description of each lettered panel in order (A, B, C…), with any sample size, statistic, or abbreviation that differs between panels stated within that panel’s description rather than only once at the end.

Should abbreviations already defined in the manuscript text be redefined in the figure legend?

Yes. A figure is frequently viewed independently of the surrounding text, so the legend needs to stand on its own — abbreviations, symbols, and color codes should be defined at their first use within the legend even if they were already spelled out earlier in the manuscript.

For the technical side of figure preparation — resolution, accepted file formats, and panel-labeling conventions across major publishers — see Scientific Figure Rules for Journal Submission. For choosing the right chart type before a figure is even drafted, see Designing Effective Scientific Figures. For image-integrity review and what counts as inappropriate processing, see Detecting Image Manipulation in Figures. For reuse of a figure that already carries someone else’s legend and copyright, see How to Get Permission to Reuse a Published Figure or Text. See also the dictionary entry for figure legend for the operational definition and additional short examples.

Last verified 2026-08-16. Journal-specific legend placement and length requirements change with individual author-guideline updates; check the target journal’s current instructions for authors before final submission.