The Discussion is the section of a research paper where authors interpret their results: what the findings mean, how they answer the research question posed in the Introduction, how they relate to existing literature, what the study’s limitations are, and what they suggest for future work. It is the one section of an IMRaD-structured paper where argument, comparison, and reasoned speculation are expected — but that latitude is easy to misuse, and a weak Discussion is one of the most common reasons reviewers request major revisions even when the underlying data are sound.
This guide covers what the Discussion is actually for, the structure most journals expect, and the pitfalls that most often draw reviewer criticism. For the section-by-section logic of the whole paper, see CASRAI’s IMRaD Structure entry; for a fully worked example of the section that precedes it, see Research Paper Abstract: A Worked Example.
What the Discussion Section Is For
The single most useful distinction to hold onto while drafting is: Results reports, Discussion interprets. The Results section is a neutral, ordered account of what the data show. The Discussion is where the author is finally permitted — and expected — to say what those findings mean, why they matter, and how confident a reader should be in that interpretation.
The International Committee of Medical Journal Editors (ICMJE) frames the whole IMRaD structure, including this division of labor, as “not an arbitrary publication format but a reflection of the process of scientific discovery.” Each section corresponds to a distinct stage of the research process; the Discussion is the stage where the author steps back from the raw findings and explains their significance. A Discussion that never goes beyond describing what was found again has not done its job, no matter how well-written the prose is.
Typical Structure of a Discussion Section
Journals and disciplines vary in exact heading conventions, and many papers run the Discussion as continuous prose without subheadings at all. But the underlying moves are consistent across most empirical papers, and drafting with these five moves in mind — in roughly this order — produces a Discussion a reviewer can follow without getting lost.
1. Summary of Key Findings
Open with a brief, plain-language restatement of the study’s main result(s) — not a repeat of every number from the Results section, but a one- or two-sentence orientation that reminds the reader what was found and directly answers the research question stated in the Introduction. This is summary, not repetition: it exists to reorient the reader before the interpretation begins, not to re-report data.
2. Comparison to Prior Literature
Situate the findings against what was already known. Do they support, extend, contradict, or complicate previous studies? This is where citations to earlier work belong — the Introduction cites prior literature to justify why the question mattered; the Discussion cites it again to show where the new findings actually land relative to that existing body of evidence. Genuinely unexpected or contradictory results deserve particular attention here: silently omitting a finding that conflicts with prior literature, rather than addressing it, is a common and avoidable weakness.
3. Limitations
Every study has constraints — sample size, study design, measurement precision, generalizability of the population studied, potential confounds that could not be controlled. State them directly and specifically, rather than in a single vague, boilerplate sentence. A limitations discussion that is specific about what the limitation actually constrains (which claims it weakens, and by how much) reads as more credible to a reviewer than one that lists limitations generically without connecting them back to the paper’s own claims.
4. Implications
Explain why the findings matter beyond the immediate result — for the field’s theoretical understanding, for methodology, or for practice/policy, as relevant to the discipline. Implications should be scaled to what the data actually support; this is the section most vulnerable to overclaiming (see below), so implications are usually strongest when they are stated with the same precision as the limitations that bound them.
5. Future Directions
Point to the specific next questions the study’s limitations and findings raise. Strong future-directions statements are concrete (a specific follow-up study design, population, or method) rather than a generic closing line like “further research is needed” — a phrase reviewers frequently flag precisely because it asserts nothing testable.
Many papers close with a short, separate Conclusion paragraph or heading that distills the Discussion into a final one- or two-sentence takeaway; whether this is a subsection of the Discussion or its own heading is journal-specific — see CASRAI’s Research Paper Format guide for how journals commonly differ on this.
Common Pitfalls
Overclaiming Beyond the Data
The most frequent substantive criticism reviewers raise about a Discussion section is that its conclusions outrun what the data can actually support. Causal language (“X causes Y”) applied to correlational or observational data, generalizing a finding from a narrow sample to a much broader population, or presenting one study as definitively settling a question are all forms of overclaiming. The fix is calibrated language: use hedges that match the actual strength of the evidence (“suggests,” “is consistent with,” “may indicate”) rather than hedges chosen out of habit, and explicitly scope claims to the population, conditions, or design actually studied.
Ignoring or Minimizing Limitations
A Discussion that omits limitations entirely, or buries a single generic sentence about them at the very end, reads to an experienced reviewer as either naive or evasive — and it is one of the more commonly reported specific issues that surfaces in reviewer comments. Limitations are not an admission that the study failed; they are what let a reader correctly calibrate how much weight to put on the findings. Naming them precisely, and explaining what claim each one actually constrains, is part of what makes a Discussion trustworthy rather than a liability that undermines it.
Simply Repeating the Results Section
A Discussion that re-walks through every result in the same order as the Results section, without adding interpretation, comparison to the literature, or implications, has failed to do the one thing the section exists for. If a sentence in the Discussion could be moved into Results without changing its meaning, it almost certainly belongs in Results instead — the Discussion should be doing interpretive work with every paragraph, not restating data that already appeared with its own tables and figures.
A Short Checklist Before Submission
- Does the opening summarize the key finding(s) without re-reporting every number from Results?
- Is every claim about what the data mean directly traceable back to a result actually reported?
- Are limitations named specifically, with a stated effect on what claims they weaken — not a single generic sentence?
- Does the Discussion engage with prior literature that disagrees with the findings, not only literature that supports them?
- Are future-directions statements concrete rather than a generic closing line?
- Could any sentence be cut-and-pasted into Results unchanged? If so, it likely belongs there, not in Discussion.
Frequently Asked Questions
What is the Discussion section of a research paper?
It is the section, following Results, where authors interpret their findings: explaining what they mean, how they relate to prior research, what the study’s limitations are, and what follow-up work they suggest. It is the interpretive counterpart to the Results section’s neutral reporting of data.
What is the difference between the Results and Discussion sections?
Results reports what was found, without interpretation; Discussion explains what those findings mean. Interpretation, comparison to the literature, and claims about implications belong exclusively in the Discussion — including a result in the Results section is a factual report, while the same content phrased as an interpretive claim belongs in the Discussion.
How long should a Discussion section be?
Length varies substantially by journal, field, and the complexity of the findings, and no single word count applies across disciplines — authors should follow the target journal’s specific guidance rather than a generic rule. As a structural matter, a Discussion is usually proportionate to how much genuine interpretation the findings support: padding it to reach a length target, rather than to cover the five structural moves above, tends to produce the repetition problem described earlier in this guide.
Do I need a separate Limitations subsection?
Some journals require an explicit “Limitations” heading; others expect limitations addressed within the flow of the Discussion without a separate heading. Either is acceptable structurally as long as limitations are addressed specifically and are not omitted — check the target journal’s author guidelines for its specific formatting expectation.
Can the Discussion include new data or results?
No. The Discussion interprets findings already presented in the Results section; introducing a new result for the first time in the Discussion, rather than reporting it in Results and then interpreting it here, is a structural error reviewers commonly flag.
For how the Discussion fits into planning and drafting a full manuscript, see CASRAI’s guides on how to write a research paper and research paper format, and the IMRaD Structure dictionary entry for how Discussion relates to the other three sections.
