What the research on rooibos actually says

I — Start here

The four questions

The biggest gapRooibos and pregnancySix papers exist. None is a trial in pregnant women. What the NHS actually says, and one widely quoted citation that turns out not to exist. The consequential oneBreastfeeding, babies and toddlersWhere the honest answer is not “probably fine”. What the WHO-co-authored study found, and why the age matters more than the tea. 6 600 searches a monthThe benefits, testedAll eight human studies, in order, with their sample sizes. What each one measured, and the three that found nothing. What is documentedSide effects and interactionsTwo liver case reports in the whole literature. The oestrogen question. What a review says about medicines.

II — The point

Why a tourism site is writing about research

Because the questions people actually ask about rooibos are not about the Cederberg. We measured it: of 1 436 distinct rooibos questions harvested from Google's own suggestion engine in September 2026, nine mentioned a place. Roughly a fifth were about pregnancy, babies or fertility. Another fifth were about benefits.

We could have kept writing about the drive from Cape Town. But a guide to the place rooibos comes from that cannot answer the questions rooibos drinkers ask is a guide with its back to its own readers.

The other reason is that this subject is badly served. The pages ranking for these questions are health-content publishers and tea shops. They contradict one another on dose. They attribute numbers to health services that those services never published. At least one reassurance they all repeat rests on a paper that does not exist.

None of that requires a medical qualification to fix. It requires someone to read the literature and write down what is in it. We can do that, and we can show our work.

What people ask

Benefits 323 of 1 436 questions
Pregnancy and babies 313
Side effects and safety 190
Caffeine 40
Hormones 36
A place or a visit 9

1 436 distinct queries harvested from Google's suggestion endpoint, 18 September 2026; method and full list in the project's research notes. Autocomplete shows the shape of demand, not its volume — do not read these as traffic figures.

III — The rules

What we will and will not write here

These pages are held to a tighter standard than the rest of the site, because the cost of getting them wrong is higher.

We report studies. We do not make claims. There is no sentence anywhere in this section saying rooibos treats, prevents, relieves or improves anything in a human being. If a study found something, we say the study found it, and we give you its size.
Every study is named, dated and linked to its record in the literature, so you can read the abstract yourself rather than take our word for it.
Sample sizes go next to findings, always. “A study found” is how eight people become a fact. Eight people are eight people.
We do not publish a dose. Not cups per day, not millilitres, not grams. We will quote a number a health service published and say whose it is.
Cells and rats are labelled as cells and rats, in the same sentence as the finding, not in a footnote.
When we cannot find something, we say we could not find it — and give you the search we ran, so you can judge whether we searched well.
Nobody here is medically qualified. Not a doctor, not a dietitian, not a pharmacologist. This section is a reading of public literature by a website, and it is not a substitute for asking someone who examined you.

Why the bar is set here

Rooibos is a food, and in South Africa the labelling and advertising regulation for foodstuffs — R146 of 2010 — prohibits calling a food “healthy” in a way that implies health-giving properties, and prohibits medicinal or therapeutic claims outright. Practitioners read it as covering websites, not only packaging.

We would hold this line anyway. But it is worth knowing if you sell rooibos: the sentence you are tempted to put on the box is probably the one the regulation names.

Regulations: Labelling and advertising of foodstuffs, GNR.146 of 1 March 2010, regulation 13, read together with the Medicines and Related Substances Act 101 of 1965. That websites fall within “advertisement” is a legal practitioner's reading, not the regulation's own words — if you are labelling a product, take advice rather than ours.

IV — The method

How we searched

Everything in this section came out of PubMed, the US National Library of Medicine's index, queried through its public programming interface on 18 September 2026. Where a page states a search, it prints the exact terms so you can repeat it.

One thing that went wrong on the first pass, since it is the sort of thing this section exists to be honest about. Our first count of the human studies said four. It was built from a relevance-sorted search, which is a reading list rather than a census, and it had missed half of them — including both of the properly controlled trials. Searching by publication type instead found eight. The figure on this page is the corrected one.

Two limits worth knowing. PubMed is not all of science — it indexes biomedical journals, so work published elsewhere, or in a language it does not cover well, can be missed. And absence of a study is not absence of an effect. When we say no trial exists, we mean we could not find one in that index on that date. That is a weaker statement than “it has been ruled out”, and we try never to let it slide into the stronger one.

If you know of a paper we have missed, tell us. A correction is cheaper than a wrong page.

The plant and the placeWhat rooibos is, why it grows only here, and how it is made. Green rooibosThe same leaf without the oxidation step, and what that does and does not change. How this site worksWho runs it, what it refuses to publish, and who pays for it. Nobody does.