No South African university sets a required number of sources for a master’s dissertation, and none of the faculty guides we searched states one. To give the question an evidence-based answer, we counted the reference lists of 25 South African master’s dissertations: the median held 75 sources, and half sat between 42 and 170.
This is one of the most-asked questions in South African postgraduate writing and one of the worst-served. Search it and you get numbers invented by content farms — “a master’s dissertation should have 50 to 100 references” — with no source, no country and no discipline attached. So we went and measured it on South African dissertations. The method, the sample and its weaknesses are all set out below, and the answer is more useful than a single number.
Is there an official minimum number of sources?
Not one that we could find. What your institution actually prescribes is length and credit weight — how many words the dissertation may run to, and how many of the degree’s credits sit in the research component. Neither of those is a source count.
Before you accept our word for it, do the ten-second check yourself: open your faculty’s higher-degrees or postgraduate guide, search it for the words references and bibliography, and read every hit. In almost every case, everything you find will be about format — which style, how to lay out the list, whether to include works you read but did not cite — and nothing about quantity. If your department is the exception and does state a number, that number overrides everything on this page.
Two things follow from the absence of a rule. First, nobody is going to fail you for having 61 references instead of 80. Second, and less comfortably, there is no number you can hit that makes the reference list “done”. The examiner is not counting. They are checking whether the literature you cite actually supports the argument you make — which is a different test entirely, and one you can fail with 200 sources.
How many sources do real South African dissertations cite?
We could not find anyone who had measured this, so we did.
The sample was drawn from OpenAlex, restricted to works recorded as dissertations and hosted in a South African university’s own institutional repository — Open UCT, ResearchSpace at UKZN, SUNScholar at Stellenbosch, UPSpace at the University of Pretoria, the Wits repository, the Unisa repository, Boloka at North-West University and the UWC electronic theses repository. That frame holds 507 documents. We drew 60 at random, downloaded the full-text PDFs, located the reference list in each and counted its entries. Degree level was read off the title page.
| Degree | Documents counted | Lowest | Lower quartile | Median | Upper quartile | Highest |
|---|---|---|---|---|---|---|
| Master’s dissertation | 25 | 13 | 42 | 75 | 170 | 567 |
| Doctoral thesis | 6 | 261 | 284 | 373 | 698 | 1 159 |
Each figure, quotable on its own:
- The median South African master’s dissertation in this sample cited 75 sources.
- Half of them cited between 42 and 170 sources. That interquartile range is four times wide — which is the real finding.
- The shortest passed reference list held 13 entries; the longest held 567.
- Doctoral theses in the same sample cited roughly five times as many sources as master’s dissertations, with a median of 373.
- Split into bands, the 25 master’s dissertations divide almost evenly: 8 cited fewer than 50 sources, 5 cited 50–99, 8 cited 100–199 and 4 cited 200 or more.
Read that last line again, because it is the answer to your question. There is no clustering around a norm. A South African master’s dissertation citing 45 sources and one citing 180 are both entirely ordinary.
How reliable is that number?
Moderately. We are publishing the weaknesses because a figure like this gets quoted, and it should be quoted with them attached.
- The sample is small. Twenty-five master’s dissertations is enough to show the spread and rule out a tight norm. It is not enough to give you a discipline-level figure, and you should not cite it as a national statistic.
- The frame is skewed by what gets indexed. OpenAlex does not index South African repositories evenly, and health sciences deposits are over-represented. Disciplines with short, article-format dissertations are therefore probably over-weighted in the low bands.
- Counting was automated, then filtered. Two detectors were run on each reference list — one for author–date entries beginning a line, one for numbered Vancouver-style entries — and the higher count kept. We then discarded any document where the characters-per-entry ratio fell outside a plausible range, which is how 60 downloads became 34 usable counts. Discarding a third of the sample is a real limitation, not a rounding error.
- We did not use OpenAlex’s own reference counts, and neither should you. Of the 507 dissertations in the frame, only 159 had any references matched by OpenAlex at all. Its referenced works field counts references it can match to an indexed record, so for dissertations it is a floor, often a very low one. This is a trap worth knowing about if you plan to do bibliometrics for your own chapter.
One broader warning came out of building this dataset, and it applies to any student running a similar query. Our first attempt filtered simply on “dissertations with a South African affiliation” and returned 1 525 records — but the set included a paper in Nature, a thesis deposited at a British university and a thesis written in Indonesian. Affiliation filters attribute work by where an author is employed, not by where the research was done or the degree awarded. Restricting to South African repositories cut the frame to 507 and made it defensible. If you build a corpus for a bibliometric chapter, check what your filter actually returned before you count it — and say in your methodology how you checked. Our guide to finding South African data covers the same discipline for national datasets.
Why is the range so wide?
Because the number is an output of four decisions you have already made, not a target you choose.
Your discipline. A quantitative health sciences dissertation testing one intervention can review its literature exhaustively in 40 sources, because the relevant trials are countable. A sociology or education dissertation engaging theory has no such boundary; it cites theorists, empirical studies, policy documents and the South African literature on the same question, and 150 is unremarkable.
Your format. Several South African universities allow a dissertation by publication, where the core is one or more journal-article manuscripts bracketed by an introduction and a synthesis. Journals cap reference lists. A dissertation built that way will look thin against a monograph-format one and be perfectly compliant.
Your method. A systematic review is its reference list, and the count runs into the hundreds by design. A small qualitative study with 12 interviews needs enough literature to frame the analysis and no more. Neither is better. If you are still deciding how big your study should be, our guide to how many participants a dissertation needs works through the same logic on the sampling side.
Your degree level. This is the one our data shows most clearly. The gap between a master’s dissertation and a doctoral thesis is not a matter of degree, it is a change of kind — the doctoral sample cited about five times as many sources. If you are unclear where your qualification sits, the difference between an honours research report, a master’s dissertation and a doctoral thesis is set out separately, as is the honours research report step by step for anyone at that level.
How do I tell whether my reference list is too thin?
Counting is the wrong test. Run these five instead — all of them are things an external examiner can see at a glance.
- Is every substantive claim in your literature review attributed? Go through the chapter and mark every sentence that asserts something about the world rather than about your study. Each one needs a source or your own data behind it. Unattributed assertions are what a thin reference list actually looks like from the outside.
- Are the seminal works in your field there? Every field has five or ten works that any examiner in it expects to see. Their absence is noticed immediately, and no volume of recent citations compensates. Ask your supervisor directly which ones they would expect — it is a fair and easily answered question.
- Is the recent literature represented? Sort your reference list by year. If the newest work is four years old, an examiner will ask what has happened since — and in a fast-moving area that alone can send a dissertation back for revision.
- Have you cited the South African literature on your topic? This is the most common gap in South African dissertations and the easiest for a local examiner to spot. If your topic has been studied here and your reference list is entirely American and European, you have a problem that adding twenty more international sources will not fix.
- Can you say, for every entry, what it contributes? Pick ten references at random and state in one sentence why each is in the list. If you cannot do it for a reference, you have padded — and padding is visible, because padded citations sit in strings of four or five behind a single generic sentence and are never mentioned again.
If several of those tests fail, the fix is reading, not searching. Our step-by-step literature review guide covers how to build coverage systematically instead of by keyword volume.
Can I have too many sources?
Yes, in two specific ways, and both are marked down.
The first is citation without engagement. A reference list of 250 attached to a literature review that summarises rather than argues reads as a reading list, not a review. Examiners describe this as descriptive rather than critical, and it is one of the most common revision requirements at master’s level.
The second is citing what you have not read. Harvesting citations out of another author’s reference list without opening the originals propagates their errors into your work, and it is detectable: the giveaway is a source cited for a claim it does not actually make. If you genuinely cannot obtain a source, cite it as secondary — “Nkosi (1998), cited in Adams (2021)” — which is honest and permitted, rather than pretending to a reading you did not do.
There is also a purely practical ceiling. Every entry must be formatted correctly and consistently in your faculty’s style, and inconsistency in a long list is exactly what gets flagged in the technical check before examination. A hundred correctly formatted references beat 300 with mixed punctuation. A reference manager makes this survivable — we compare the reference managers worth using in South Africa, including the free ones — and if your list is already out of alignment with your faculty guide, there is a repair procedure for that. Which style you are in is a separate question, covered in Harvard versus APA at South African universities.
Frequently asked questions
How many sources should a master’s dissertation cite in South Africa?
No South African university prescribes a number. In a 2026 sample of 25 South African master’s dissertations, the median reference list held 75 sources and half held between 42 and 170. Treat that as a sanity band rather than a target, because the correct number depends on your discipline, method and dissertation format.
Is 50 references enough for a master’s dissertation?
It can be. Eight of the 25 South African master’s dissertations we counted cited fewer than 50 sources. Fifty is adequate if it covers the seminal works, the recent literature and the South African work on your topic, and thin if it does not — the count is not what decides it.
How many references does a doctoral thesis need?
Substantially more. The six South African doctoral theses in our sample cited a median of 373 sources, ranging from 261 to 1 159 — roughly five times the master’s median. A doctoral literature review is expected to demonstrate command of an entire field rather than of one question within it.
Does my university have a minimum number of references?
Almost certainly not, but check rather than assume. Search your faculty’s higher-degrees guide for “references” and “bibliography”. What you will normally find are rules about referencing style and reference-list format, not about quantity. If a number is stated anywhere, it applies to you and overrides any general guidance.
How many sources should an honours research report cite?
Fewer, in proportion to its scale. An honours research report is a much smaller piece of work than a master’s dissertation, with a correspondingly smaller literature review. No number is prescribed there either; the same coverage test applies at a smaller scale.
Do I count sources or reference-list entries?
Entries. One work equals one entry regardless of how many times you cite it in the text. If your faculty guide asks for a bibliography rather than a reference list, the count may also include works you consulted but did not cite — check which of the two your guide requires, because they are different documents.
Can I include sources I read but did not cite?
In a reference list, no — a reference list contains only works cited in the text. In a bibliography, yes. Most South African faculty guides ask for a reference list, so uncited works have to come out. Padding a reference list with uncited reading is a technical error and easy to detect.
Do websites, policy documents and theses count as sources?
Yes. Government publications, institutional policy documents, statistical releases and other dissertations are all legitimate entries and all belong in the count. Some disciplines expect a majority of peer-reviewed journal articles; grey literature is often exactly the right source for a South African policy or context chapter.
How old is too old for a source?
There is no cut-off, and mechanical rules like “nothing older than five years” cause more harm than good — they push students to drop the foundational works of their own field. The workable test is different for the two categories: theoretical and seminal works can be any age, while empirical work on the current state of your topic should be recent enough that an examiner cannot name something newer that you have missed.
Does AI-assisted writing change how many sources I need?
No, and it introduces a specific risk. Language models generate plausible-looking citations that do not exist, and a fabricated reference in a submitted dissertation is a serious matter. Verify every reference against the actual source before it enters your list. What South African universities permit varies by institution and is set out in our guide to whether AI is allowed for dissertations here.
Will an examiner actually count my references?
No. External examiners assess whether the literature you cite supports your argument, whether you have missed anything an expert would expect, and whether you engage with sources rather than list them. Our explanation of how external examination works in South Africa covers what they are actually asked to report on.
Where can I check the numbers in this article?
The frame is OpenAlex, filtered to works of type dissertation hosted in the eight South African institutional repositories named above, giving 507 documents; 60 were sampled at random and 34 produced a reference count that passed the plausibility filter, of which 25 were master’s dissertations and 6 doctoral. Counting was automated from the full-text PDFs. Treat the figures as indicative of the spread, not as national statistics.
Stop counting and start covering
The question “how many sources?” is nearly always standing in for a harder one: have I read enough? Our data answers the first question — the median South African master’s dissertation cites 75 sources, and anything from 42 to 170 is thoroughly normal — but it deliberately cannot answer the second, because the second depends on a field you know better than we do. The five coverage tests above answer it far better than any count.
What we can do is remove the mechanical part. Tesify helps you build and format a reference list that matches your faculty guide, keeps every citation attached to a source you actually read, and leaves the argument yours to make. If you are writing around a job and a deadline, the system for writing a dissertation while working full-time is where to start.