The question comes up in every bachelor’s thesis seminar, usually around week three: how many sources does my thesis actually need? The honest answer is: nobody knows exactly — it depends. The useful answer is a framework of rules of thumb that you calibrate against your field and your method. This post gives you both: concrete numbers and the factors that move them.
The short answer
For the impatient — benchmarks that hold across most fields:
| Thesis | Length | Typical source count |
|---|---|---|
| Seminar paper | 10–20 pp. | 10–20 |
| Bachelor’s thesis | 30–60 pp. | 25–50 |
| Master’s thesis | 60–100 pp. | 50–100 |
| Doctoral dissertation | 150+ pp. | 150–400+ |
These numbers aren’t exam rules — they’re what examiners see in practice and accept as “normal”. Land far below, and the work looks under-researched; land far above, and it often looks nervous. Both stand out, and in both directions you can defend a departure — but you need a reason.
The rule of thumb: one source per body page
If you have no time for long comparisons: one to one-and-a-half sources per body page is the benchmark that holds across most fields. “Body page” means: excluding title page, table of contents, appendix, reference list — only the actual chapters.
- 30 body pages → 30–45 sources
- 40 body pages → 40–60 sources
- 60 body pages → 60–90 sources
That’s a starting number. What pushes it up or down comes next.
What drives the number up
Theoretical, literature-based work. If your thesis is essentially a literature review or sets several theoretical positions against each other, you need more sources — simply because every position has to be represented. A humanities bachelor’s thesis on “Nietzsche’s influence on Foucault” easily lands at 50–80 sources.
Systematic literature review. Rules of thumb stop applying here — the number falls out of your search strategy. 100+ included studies is normal, plus methods sources. If you follow PRISMA, the number is the result of your screening, not your choice.
Humanities and social sciences. History, literature, philosophy, sociology cite more densely than engineering — not because the field requires it, but because arguments in these fields are supported by other arguments, not by measurements. Bachelor’s theses with 60–80 sources are not unusual.
Interdisciplinary topics. Anyone writing on AI ethics in medicine has to cite from at least three fields (computer science, ethics, medicine) — automatically more sources than a disciplinarily “clean” thesis.
What drives the number down
Empirical theses with your own data collection. A bachelor’s thesis where you conduct 20 interviews yourself, analyse a survey, or run an experiment has a substantial methods and results section that needs no citations. 20–30 sources is entirely usual here — the contribution lies in your data, not in reading secondary literature.
Technical and mathematical fields. In mechanical engineering, computer science, or physics it’s normal for large sections (derivations, proofs, code, test results) to be uncited. The literature section can stay compact — 15–25 sources, often just standard works plus the two or three relevant recent papers, is normal here.
Applied and project theses. A bachelor’s thesis that solves a practical problem in a company (“Introducing a CRM system at company X”) leans on a few core sources plus internal documentation. 15–25 sources is often enough — provided those sources carry the theoretical grounding credibly.
What actually counts as “one source”?
A source is an entry in the reference list. Sounds trivial, but people miscount all the time:
- A book you cite ten times → one source.
- Ten different chapters from an edited volume → ten sources (each chapter gets its own entry because the authors differ).
- Two editions of the same work → two sources (they have different years, often different content).
- An interview you conducted yourself → goes into the reference list, but many exam regulations treat it as “primary data” rather than a classical source. Ask.
The key thing: what counts is not the number of in-text citations, but the number of entries in the reference list. Confusing the two produces either ridiculously high or suspiciously low numbers.
Quality beats quantity — and examiners see the difference
Fifty sources from random websites are worse than twenty from peer-reviewed journals. What examiners check when they open the reference list (before reading a word of your thesis):
Scholarly-to-popular ratio. In most fields the expectation is 70–90% peer-reviewed or clearly scholarly sources. “Clearly scholarly” means: academic books from established publishers, journal articles with a review process, dissertations, conference proceedings. Everything else is grey literature or a grey area.
Recency. In empirical fields, at least 60% of sources should be from the last ten years. Submit a 2026 bachelor’s thesis on machine learning with no source after 2020 and you look careless. In the humanities, age is often a mark of quality — the classics belong in the list.
Language. A German-only bibliography in a field with English-language research (nearly all natural sciences, business, psychology) is a red flag. Conversely: English-only citations in a comparative-law topic that treats German law is a red flag too.
Publisher/journal spread. If twelve of your twenty sources come from the same edited volume, you wrote a book report, not a bachelor’s thesis. Same if everything comes from one author or one journal.
How to estimate your target number
A simple procedure that works in practice:
- Write down your chapter outline. Ten chapters/sub-chapters is normal for a bachelor’s thesis.
- Per chapter, estimate: how many sources do I need to back up the claims? An introduction needs 3–5, a fundamentals chapter 8–12, a methods section 3–5 (only methods literature), a discussion 5–10 (because you place your results in the literature).
- Add it up. That’s your rough target.
- Cross-check with the rule of thumb. Does it fit the frame? If you land at 15, that’s too few; at 120, too many for a bachelor’s.
The number isn’t set in stone. It shifts during writing — that’s normal.
Red flags examiners spot immediately in a reference list
- All entries from the same publication year. Reads like a copy-paste from one review article.
- Conspicuously many “forthcoming” or “in preparation” entries. These citations are rarely verifiable — examiners go after them on purpose.
- Only websites, no books or journals. For most topics, not a plausible research result.
- Absurdly many self-citations of the supervisor. Stands out instantly and reads as tactical.
- Dead DOI or 404 URL. One of the first things examiners spot-check by clicking.
These are signals that stand out without anyone reading the thesis itself. Avoid them and you’ve cleared a big chunk of the initial scepticism.
Too few or too many: what to do
Too few: Usually you’re missing foundational literature (the standard work on your topic, a classical overview article) or methods literature. Check that every central concept is defined by at least one cited source, and every method has the handbook or paper that explains it.
Too many: The most common cause is “defensive citing” — every claim gets three references when one would do. Trim to the single most precise reference per claim. Anything that only repeats a concept instead of extending it can go. After this pass, bachelor’s theses often shrink from 90 to 55 sources without losing an argument.
And the biggest misconception: the number doesn’t protect you
Anyone who believes many sources automatically make a thesis better underestimates what examiners actually check. What matters is not whether you have sixty or thirty sources — it’s whether the claims you attribute to your sources are actually there. Misattributed citations, wrong page numbers, hallucinated AI references: those are the mistakes that cost you a grade, and they occur independent of source count.
That’s exactly where Acurio comes in. You upload your thesis and your source PDFs, and Acurio checks every single citation for content against the source. Does the page number match? Does author X really say what you claim? Does the source exist at all? That’s the part neither a plagiarism scanner nor a reference manager can cover.
For a quick check you don’t even need to upload anything: the free Quick Check analyses your reference list in the browser, without signup, and flags problematic entries in seconds — unfindable sources, contradictory metadata, dead DOIs.
The takeaway
Count your sources by the rule of thumb — one to one-and-a-half per body page — and calibrate against your field and method. Empirical means fewer, theoretical means more. The exact number matters less than the scholarly-to-popular ratio, recency, publisher spread — and whether every individual entry holds. Examiners rarely count entries. They check whether the list is sound and whether the in-text citations deliver what they promise.