How To Write A Figure Caption Biology

Writing a Figure Caption That Actually Gets Accepted

Most biology papers get desk rejected or sent back for revision because the figure captions are sloppy. It happens all the time. The reviewers and editors read them before anything else. They are the first thing anyone looks at after the title. Getting them right matters more than most people realize.

A figure caption in biology has a very specific job. It needs to tell the reader what they are looking at without making them flip back and forth between the text and the image. The standard structure covers the biological material, the experimental condition, what each panel shows, and the statistical treatment. That is the baseline. Anything less than that leaves the reader guessing. Start with a short descriptive phrase that identifies the experiment or observation. Then move into the details. I usually write these in a single paragraph that runs between 80 and 150 words. Here is the order I follow every time: what the sample is, what was done to it, what each panel represents, what the error bars mean, and what the statistical test was. If there are abbreviations or gene names, define them the first time they appear in the caption. Do not assume the reader remembers the Methods section. Let me give you a concrete example from my own work. I was writing a paper on mouse tumor growth curves and had six panels per figure. The reviewer flagged that I had written "Treatment reduced tumor volume compared to control" without specifying the time point, the sample size, or the statistical test. I rewrote that single line to read: "Treatment with compound X (10 mg/kg, i.p., daily) significantly reduced tumor volume compared to vehicle control at day 21 (n = 12 per group, two-tailed unpaired t-test, p = 0.003)." The change took thirty seconds. It prevented the reviewer from asking the same question three times during the rebuttal.

The abbreviations are where most people mess up. If you use "ns" for not significant, make sure the journal allows it. Some journals require you to write out "not significant" in full. If you use "SEM" or "SD," pick one and be consistent across every figure in the paper. I have seen drafts where panel A uses SD and panel B uses SEM, which makes the reviewer think you do not understand the difference. It looks careless even if it was just a copy-paste error. There is a nuance that beginners consistently miss. You should never repeat information in the caption that is already completely described in the figure legend or panel label. But you also should not assume the figure is self-explanatory. The caption and the figure work together, not independently. If the figure shows a Western blot, the caption needs to state the antibody, the molecular weight marker, and how quantification was performed. The image itself will show the bands. The caption explains why those bands matter. I encountered a real edge case once that I still think about. We had a time-course experiment with fifteen time points across four panels, and the journal required us to label every individual data point rather than just showing mean bars. My original caption described the experiment broadly but did not account for the individual point labeling requirement. I had to go back and add a sentence specifying that each symbol represents an individual biological replicate and reformat the graph in GraphPad Prism to display the points. That took about forty minutes total. The alternative would have been a major revision cycle that could have added two weeks to the timeline.

Statistical reporting in captions has become stricter over the last few years. Simply writing "p

0.05" is no longer acceptable in most journals. You need the exact p-value to two or three decimal places, the test used, the degrees of freedom where relevant, and the correction method if you ran multiple comparisons. Bonferroni, Tukey, Holm-Sidak—name the correction and state it in the caption or the Methods, but pick one and stick with it. Another thing people overlook: figure panel labels. A and B, not A and C. It sounds trivial but I have personally had figures returned by the editorial office because the panels were mislabeled in the caption relative to the image. One of my early papers went back for exactly that reason. The figure showed panels labeled A through D, but the caption referred to panels A, B, C, and E. The editor did not flag the content at all. It was purely a labeling mismatch. I fixed it in twenty minutes. It cost me three days of turnaround time because the manuscript had to re-enter the queue. If you are working with microscopy images, the caption must include the magnification, the scale bar description, the fluorophores or stains used, and the image processing steps. Did you apply a flat-field correction? Was there deconvolution? Did you adjust brightness or contrast globally or per channel? Journals now require you to state whether any processing was applied and whether it was applied uniformly. If you did not process the images at all, say that explicitly. It matters.

Here is the part that tends to annoy people, and it is worth stating plainly: figure captions in biology are one of the harder things to get right when you are juggling multiple experiments. They are easy to treat as an afterthought because you spend most of your time generating the data. But a poorly written caption can undermine a perfectly good result. Reviewers interpret vague captions as vague thinking. They are not always fair about it, but they do it anyway. The biggest bottleneck I see is people writing captions before the figures are finalized. Drafting a caption and then realizing the figure changed means rewriting the entire thing. I recommend finalizing the figures first, then writing the captions in one focused sitting. It takes me about twenty minutes per figure if the data is clean and the statistical tests are straightforward. If I am dealing with a complex multi-panel figure with multiple conditions, closer to forty-five minutes. Never longer than an hour. If it takes longer, you are probably overthinking it or the figure itself needs reorganization. For the actual format, keep it as one continuous paragraph. Do not use bullet points inside the caption unless the journal specifically allows it. Some open-access journals do, but most traditional subscriptions do not. Follow the author guidelines for the target journal exactly. They vary enough that assuming one format works across the board will get you a correction notice before peer review even starts.

If you are submitting to a high-impact journal, expect the captions to be scrutinized more heavily. Nature, Cell, and their companion journals often require additional details like sample size justification, blinding statements, and reproducibility notes within the caption itself. That is not extra fluff. It is a standard requirement now. Skipping it will slow down your submission. The workaround I use for managing captions across a full paper is a simple spreadsheet. Each row is a figure, each column tracks the panel labels, the biological material, the statistical test, the p-values, the abbreviations used, and whether the journal requires a specific statement. It sounds over-engineered but it catches inconsistencies that your brain will miss on the tenth figure of the twelfth draft. I used it on a nine-figure paper last year and caught three mismatches between caption text and actual panel labels that would have otherwise gone to review. Finally, remember that a caption is a standalone document. Someone should be able to read it without reading the main text and understand what the figure communicates. That is the metric that matters. If you can achieve that in under 150 words with all the necessary details included, you are doing it correctly. Anything longer and you are either including irrelevant information or the figure itself is too complex and needs to be split.