Sumatran coffee is known for wet hulling, which Indonesians call giling basah: the parchment comes off while the bean is still wet, at 40 to 45 percent moisture in a 2005 Indonesian trial, rather than after drying to about 12 percent. That trial’s authors say the handling “presumably contributes” to Mandheling coffee’s unique cup character.
The study (Ismayadi, Marsh and Clarke, Pelita Perkebunan 2005), in the Indonesian Coffee and Cocoa Research Institute’s own journal, sets out the whole sequence: “following de-pulping and 1-2 days sundrying, wet parchment is stored for varying periods up to a few weeks, the parchments are then de-hulled when still wet (40-45% moisture content) then the beans sundried.”
The same paper describes the resulting cup without any help from us: Mandheling coffee “is characterised by low acidity, heavy-complex body, spicy-little earthy and fruity flavor”, produced “by smallholder farmers in the highland surrounding Lake Toba North Sumatra”. Note the paper’s hedge: the handling procedure “presumably contributes to the unique cup character”. The paper does not prove the causal chain; the association is long-observed and the method is genuinely unusual.
Table of Contents
What wet hulling actually does, and the risk that comes with it
Normal processing dries parchment coffee down to around 12 per cent moisture before hulling. Indonesia’s own national standard for green coffee beans, SNI 01-2907-2008, sets 12 per cent as the reference moisture content, and researchers working on Gayo arabica hold their samples there for exactly that reason. Wet hulling breaks that sequence: the bean is stripped of its parchment while still soft and swollen, then dried naked in the sun. Handling green coffee at 30 to 45 percent moisture is thought to give wet-hulled beans their unusual appearance and, the trade holds, their heavy body and low acidity.
Sources differ on the moisture at hulling: 40 to 45 percent in the study above, and “between 30 and 35%” in Wikipedia’s Giling Basah article. The article’s only citation for that passage is a 2006 UNDP review of the Aceh coffee industry by T. Marsh, which we have not opened.
Storing wet parchment for days or weeks before hulling has an obvious hazard, and the study measured it rather than speculating. Its design used the normal wet process, dried thoroughly to 12 per cent, as the control, and stored wet parchment in polyethylene sacks for 1, 7 and 14 days. The findings were mixed and are worth reporting in full rather than selectively:
- Moulds did grow. “black Aspergillus was the dominant species found in the beans, while A. ochraceus an OTA producer, was found in some samples with low infection rate (0-15.3%)” — the missing comma is the paper’s own.
- The toxin mostly did not follow. “OTA was found in only 5 samples out of 42 samples with range of 0.17-2.24 ppb, very less than European Union limit.”
- Length of storage did not predict the outcome. “There was no clear trend of storage period on the mould infection rates, OTA content, and the Mandheling cup characters”, which the authors attribute to the variability of the farmer lots they used.
- The existing local practice helps. “Spreading of coffee inside the warehouse during the day could suppress moulds growth.”
In short, wet hulling creates a real mould pathway, and in this trial the ochratoxin A it found was, in the paper’s words, “very less than European Union limit”, a limit the abstract does not state. One trial on six lots is not a food-safety clearance for an entire origin, and we are not presenting it as one.
Wet-hulled green beans are often said to have a distinctive blue-green cast. Wikipedia’s Giling Basah article says the process “gives the beans a unique bluish-green appearance”, citing the same 2006 UNDP review; no study we opened measured it, so this page reports the claim rather than asserting it.
Indonesia is a robusta country with a famous arabica sideline
This is the piece of context the old page never gave. On the USDA Foreign Agricultural Service forecast for 2026/27, Indonesia produces 11.38 million 60-kg bags — 10.0 million of it robusta and 1.38 million arabica. Roughly seven bags in eight are robusta, Coffea canephora. The Sumatran arabica that carries the country’s reputation abroad is a small fraction of what the country actually grows.
| Indonesia, thousand 60-kg bags | 2021/22 | 2024/25 | 2026/27 forecast |
|---|---|---|---|
| Robusta | 9,300 | 9,300 | 10,000 |
| Arabica | 1,280 | 1,400 | 1,380 |
| Total | 10,580 | 10,700 | 11,380 |
From USDA FAS, Coffee: World Markets and Trade, July 2026. Most of the robusta is lowland: USDA forecasts Indonesia’s 2026/27 crop at 11.38 million bags, down from 12.37 million in 2025/26, with all of the fall in robusta (11.0 to 10.0 million), because “excessive rainfall disrupted flowering and cherry formation in the lowland areas of Southern Sumatra and Java, where approximately 75 percent is grown”, a share USDA gives in its robusta paragraph without naming the crop. Gayo in Aceh, Lintong and the Mandheling highlands sit above all that, and are where the wet-hulled arabica comes from.
The USDA figures above are from USDA’s July 2026 release; its next world coffee report is due on December 16, 2026.
What does the research say about kopi luwak?
Civet coffee is the thing about Indonesian coffee that most readers arrive already knowing, so it should be handled properly rather than sold.
The animal welfare evidence is specific, recent and bad. Lewis-Whelan and colleagues assessed 29 civet coffee tourism plantations in Bali and eight wildlife markets that supply them, publishing in the Journal of Applied Animal Welfare Science in 2024. They “found 99 civets in civet coffee tourism plantations and 136 in markets”, estimated that “annually between 400 and 800 civets are sold”, and reported flatly that “welfare standards do not meet the Indonesian laws and regulations” and that “no civet met all of the five freedoms”. Diet, hygiene, water availability and mobility all affected the animals’ body condition scores. Their recommendation is that tourists “should be made aware of the low welfare standards to inform their choice about supporting this industry”.
Much of what is sold as kopi luwak is not kopi luwak. Adulteration is a well-enough recognised problem that analytical chemists have built methods to detect it — Suhandy and Yulia published a UV-visible spectroscopy and partial-least-squares approach in the International Journal of Food Science in 2017 specifically “to protect the authenticity of luwak coffee and protect consumer from luwak coffee adulteration”. Earlier metabolomics work by Jumhawan and colleagues did the same job with discriminant markers. Nobody builds an authentication method for a product that is not being faked.
A widely repeated figure says 42 per cent of kopi luwak is fake or adulterated. It appears in a 2021 commentary in Foods by Lachenmeier and Schwarz, who write that “it is almost a relief that much coffee labelled as Kopi Luwak is probably a counterfeited product that has never seen the digestive tract of an animal”, and put the 42 per cent in brackets as a claim. We followed their citation: it leads to a 2007 report by National Public Radio, not to a study. Treat the general point as sound and the specific percentage as unmeasured.
That same commentary makes a taxonomic argument worth flagging: examining bean shapes in the published photographs of “digested” coffees, the authors conclude that “the actual coffee species under investigation has been Coffea liberica, not Coffea arabica or Coffea canephora, which has been incorrectly assumed”. If they are right, some of the chemistry literature on kopi luwak has been comparing the wrong species. Our page on liberica covers that species, including the 2025 genomic work by Davis and colleagues that split it into three — C. liberica, C. dewevrei (excelsa) and C. klainei.
We are not recommending kopi luwak and we do not link to sellers of it.
Kopi tubruk, and the rest of the archipelago
Two things the old page got right and are kept. Kopi tubruk is a real and specific preparation — grounds boiled with sugar and served unfiltered, with the grounds settling in the cup — and street coffee stalls, warung kopi, are a genuine part of daily life rather than a travel-brochure flourish. Java’s name became a synonym for coffee because Dutch cultivation started there. Sulawesi’s Toraja highlands use wet hulling too, which is why Sulawesi and Sumatra are so often described in similar terms.
For the species themselves, see our pages on arabica and robusta; for how a warming climate is expected to redraw the growing map in Southeast Asia, see climate change and coffee farming. For Indonesia’s place in the wider region — a region whose species-split crop is 89 percent robusta, where Sumatra’s famous wet-hulled arabica is a small fraction of what the country actually grows — see our Asian coffee overview.
Sources
- “Influence of Storage of Wet Arabica Parchment Prior to Wet Hulling on Moulds Development, Ochratoxin A Contamination, and Cup Quality of Mandheling Coffee”, Pelita Perkebunan (a Coffee and Cocoa Research Journal) 21(2), 2005, published by the Indonesian Coffee and Cocoa Research Institute.
- Wikipedia, Giling Basah: the 30 to 35 percent figure and the bluish-green appearance; it cites a 2006 UNDP review by T. Marsh, which we have not opened.
- Lewis-Whelan, Ardiansyah, Roberts, Nijman, Damianou, Morcatty, Birot, Imron and Nekaris, “Welfare and Management of Civets in Civet Coffee Tourism Plantations”, Journal of Applied Animal Welfare Science 27(3), 2024, 561-574.
- Suhandy and Yulia, “The Use of Partial Least Square Regression and Spectral Data in UV-Visible Region for Quantification of Adulteration in Indonesian Palm Civet Coffee”, International Journal of Food Science, 2017.
- Lachenmeier and Schwarz, “Digested Civet Coffee Beans (Kopi Luwak) – An Unfortunate Trend in Specialty Coffee Caused by Mislabeling of Coffea liberica?”, Foods 10(6):1329, 2021.
- Yulia, Analianasari and colleagues, “The Authentication of Gayo Arabica Green Coffee Beans with Different Cherry Processing Methods”, Foods 12(23):4302, 2023 — source for the SNI 01-2907-2008 reference moisture content.
- USDA Foreign Agricultural Service, Coffee: World Markets and Trade, July 2026.
Corrections to earlier versions of this page
The moisture figure was stated twice without a source. The page said the parchment is hulled “when the moisture content reaches around 30%-40%”, in the body and again in the FAQ block. The Indonesian Coffee and Cocoa Research Institute study puts the practice at 40 to 45 per cent, and the wet parchments it actually sampled from farmers and collectors in the region measured 41.74 to 53.96 per cent moisture. Wikipedia’s Giling Basah article gives 30 to 35 percent, so the sources differ; the old range of 30 to 40 percent overlapped Wikipedia’s and stopped where the study’s begins.
The source for giling basah was a YouTube video. The word was linked to a video rather than to any study or industry document. The page now cites and links the paper instead.
Three tasting-note tables have been removed. Sumatra as “earthy, spicy, herbaceous”, Java as “sweet herbal notes, full-bodied”, Sulawesi as “bold, dark chocolate, ripe fruit, subtle earthiness”, plus a variety table giving Kopi Luwak “smooth, rich, less acidic” and Mandheling “hints of chocolate and tropical fruit”. Nobody at this site has cupped any of these coffees. The one cup description on this page now is quoted from a published study and attributed to it.
Kopi luwak was described twice in appealing terms and qualified only in passing. The page said the civet’s “digestive enzymes alter the structure of the coffee beans, resulting in a smoother and less acidic brew” and that the process produces “a coffee that’s smooth, rich, and less acidic”, with no source for either claim, followed by a single sentence about “questionable animal welfare practices”. That balance is the wrong way round, and the section above replaces it.
The FAQ block and the closing “Conclusion” section are gone, and their one piece of genuine content — the processing answer — is in the opening paragraphs where it belongs.
Also corrected: an earlier version called the old 30 to 40 percent moisture figure “too low” on the strength of one study, when another cited source gives 30 to 35 percent; it said we did not find the blue-green cast in any study or standard we opened, and Wikipedia’s article states it and cites a 2006 UNDP review; and it put the 75 percent lowland share on the whole crop rather than on the robusta.
