Aged Sake: What Koshu Is and What Happens in the Bottle

Yes, sake ages, and the deliberate version of it has a name. Koshu 古酒 (literally “old sake”) is sake a brewery has matured on purpose before shipping it: it turns amber, loses its fruit, and picks up caramel, nuts and soy sauce. But the bottle that has been standing next to your kettle since 2019 is not koshu. It is something chemically different, and once you know which compounds do what, you can tell the two apart by smell alone.

What koshu means, and what the label is actually allowed to say

Japan’s National Research Institute of Brewing (NRIB), the government brewing laboratory, defines the category in one clause: sake “aged intentionally before shipping”, showing “aromas reminiscent of caramel, nuts, spice”. The operative word is intentionally. Koshu is a production decision, not a description of how long a bottle has existed.

Here is the part that surprises most people: there is no legal definition of koshu. Japan’s Standards for Manufacturing Methods and Quality Labelling of Sake (National Tax Agency Notification No. 8 of 1989, amended since) is the rulebook that defines junmai, ginjo, honjozo and the rest — and it contains no koshu category at all.

What the standard does regulate is a number. 貯蔵年数 (chozo-nensu), “years in storage”, appears on the list of optional label items, alongside things like 樽酒 (taruzake) and 生酒 (namazake). The rule is precise: it means the years counted from the day after the sake went into the storage vessel to the day storage ended, with any fraction of a year rounded down. And if a bottle blends sakes of different ages, the brewery must print the shortest one, not an average and not the oldest.

So “koshu” on a bottle is the brewery’s word; the years figure, if they print one, is the state’s. That is the reverse of what a wine drinker expects, where the vintage is the regulated fact and the house style is free text.

The Maillard reaction, running for years instead of minutes

Sake finishes fermentation carrying a lot of free amino acids and some residual sugar. Put amino acids and sugars in the same liquid and give them time, and you get the Maillard reaction — the same family of chemistry that browns toast and crusts a seared steak. The difference is the conditions. In a pan it runs in minutes at 150 °C or more. In a bottle of sake it runs for years at cellar temperature.

It produces two things worth caring about. Color, first: brown nitrogenous polymers called melanoidins, which is why aged sake shifts from near-colorless through straw to amber. Then aroma, where one compound dominates.

Sotolon is the headline. NRIB’s flavor reference data gives it a recognition threshold of 2.3 µg/L and a range across sake of 0–140 µg/L, and lists its significance flatly as the “major flavor impact character of aged sake”. The descriptors clustered around it are caramel, honey, dried fruit, molasses and soy sauce. NRIB gives two formation routes: the Maillard reaction, and a condensation between α-ketobutyric acid and acetaldehyde.

Researchers at NRIB put numbers on it. Working with sake up to 26 years old, they found sotolon present well above its recognition threshold — the odor unit in the 26-year-old sample was 61, meaning sixty-one times the concentration at which a trained nose registers it (Isogai et al., 2004). Furfural, another compound that rises with age, stayed below its own threshold. So the two compounds both accumulate, but only one of them is doing the smelling.

If sotolon sounds familiar, it should: it is a major contributor to the character of aged Sherry and of vin jaune from the Jura. That is not borrowing or imitation. It is the same molecule arriving by a similar route in a different liquid — two traditions that never consulted each other landing on the same compound, because the chemistry was always available to both.

The other thing that happens to old bottles

Sake also has a recognized fault of age, and it is a completely separate reaction. It is called 老ね香 (hine-ka). NRIB describes it as stale, burnt and sulfurous, and the sensory reference points are cooked onion, rotten cabbage and takuan pickled radish.

The chemistry behind it is sulfur, not sugar. The main culprits are dimethyl trisulfide (DMTS) and isovaleraldehyde, formed by the degradation of amino acids rather than their reaction with sugars. In 2009, NRIB researchers identified a previously unreported precursor in sake — 1,2-dihydroxy-5-(methylsulfinyl)pentan-3-one, which they named DMTS-P1 — and showed that doubling its concentration in a sample doubled the DMTS produced during storage (NRIB, 2009).

The most telling detail is how NRIB generates hine-ka for laboratory work: four weeks at 45 °C. That is the whole argument in one line. Hine-ka is not a function of years. It is a function of heat and oxygen, and you can manufacture it in a month.

Two bottles can both “taste old” and be chemically unrelated. Maturation is a slow amino-acid-and-sugar reaction that builds sotolon and color. Hine-ka is heat-and-oxygen damage that builds sulfur compounds. Which means a brewery making koshu is not simply waiting — it is running the first reaction while starving the second, and the main lever it has is temperature. That is why koshu is a category rather than a fault, and it is also why the bottle you forgot above the fridge is neither one. It did not become koshu. It just got warm.

A third thing, which is not aging at all

There is one more way a bottle goes wrong, and it gets misfiled as age constantly. Ultraviolet light produces 日光臭 (nikko-shu), light-struck odor, which NRIB describes as an unpleasant onion-like smell. UV also breaks down amino acids directly, which adds bitterness.

Two things make this worse than people assume. The damaging region is light below 400 nm, and fluorescent tubes emit it as well as sunlight — so a bottle can be light-damaged on a shop shelf, having never seen a window. And it is fast: days and weeks, not decades. The brown and dark green glass sake shares with wine and beer is a UV filter rather than a styling choice, and the paper wrap some breweries use does the same job. Nikko-shu also smells of onion, which is precisely why it gets misread as “it must be old”.

What the brewery is actually controlling

NRIB names three enemies: temperature, light and oxygen. Their published storage guidance splits sake in two. Long-aged sake and junmai-shu keep in a cool place at 5–15 °C (40–60 °F), roughly wine-cellar conditions. Namazake, sparkling sake and the whole ginjo family need 4 °C (40 °F) or lower, because their aromatics are the most fragile thing in the room. NRIB characterises long-aged sake as sake properly stored for ten to twenty years.

Temperature is the dial, and it cuts both ways: warmer storage reaches color and caramel sooner, but runs the sulfur chemistry faster too. A brewery picking a maturation temperature is choosing where to sit between those two, which is why koshu houses differ so much.

On oxygen, NRIB is measured: sake resists it better than wine, but is not impervious. Seal tightly, finish within a month, and always store the bottle upright — laying bottles down is a habit built around keeping a cork wet, and it does not transfer. We covered what happens after opening in how long sake lasts after opening, and the pasteurization question underneath it in what nama means on a label.

Should you age a bottle yourself?

Probably not, and the reason is not that it is forbidden. It is that a normal home offers the wrong conditions with some precision. Cupboards swing several degrees between day and night, kitchens run warm, and the reaction you want is the slow one — so a bottle left in a cupboard is being nudged toward the sulfur pathway rather than the caramel one. Ginjo and daiginjo are the worst candidates of all, because the delicate ginjo-ka you paid for is the first thing heat takes away.

None of which makes koshu inaccessible. It makes it something to buy rather than to attempt: a brewery that aged sake deliberately held it at a chosen temperature, in the dark, for a decade or more, and that is infrastructure rather than patience.

---

Frequently asked questions

How old does sake have to be before it counts as koshu?

There is no legal minimum, because there is no legal category. NRIB describes long-aged sake as sake properly stored for ten to twenty years, and industry usage is looser than that, often starting around three. The only figure with a rule attached is 貯蔵年数, years in storage, and printing it is optional.

My sake has turned yellow. Is it ruined?

Color on its own is not a verdict. Amber is what the Maillard reaction does, and it happens in a well-kept bottle and a badly-kept one alike. Your nose settles it: caramel, honey, dried fruit and nuts mean maturation; cooked onion, cabbage or pickles mean hine-ka or light damage. The sulfur note is unmistakable once you have met it.

Can I just keep sake in the fridge to age it?

You can keep it in the fridge to preserve it, and for nama, sparkling and ginjo styles that is exactly right. But cold slows every reaction, including the one that makes koshu interesting. A fridge is a pause button, not a maturation chamber.

Does koshu taste like Sherry?

They share sotolon, so the caramel-and-walnut register really does overlap. What does not carry over is structure: sake has no tannin and far less acid, so koshu lands rounder and softer, without the cut that pulls an oloroso back. Close enough to be a useful bridge, different enough to taste side by side.

What do you drink koshu with?

It behaves less like a table sake and more like a fortified wine, so it handles what defeats lighter styles: aged cheese, soy-braised meat, dried fruit, dark chocolate. The amino acids that drove the color change also carry a lot of savouriness — the mechanism we unpicked in umami in sake — and serving temperature matters more here than usual, which our temperature guide covers.

---

Written by Dr. Sake. We take sake apart — the chemistry, the history, the label — and put it back together in plain English. For weekly diagrams, label breakdowns and brewery notes, follow @thesakeanatomy on Instagram.

Sources

• National Research Institute of Brewing (Japan). Japanese Sake Essentials — Preservation — temperature bands, the definition of long-aged sake, and the flavor reference entries for sotolon (threshold 2.3 µg/L, range 0–140 µg/L) and hine-ka (four weeks at 45 °C; isovaleraldehyde and DMTS). NRIB, English
• National Research Institute of Brewing (Japan). Handy Storage Guide to Sake — nikko-shu, the sub-400 nm UV region, bottle color and light transmission, upright storage. PDF, English
• Isogai, A., Utsunomiya, H. and Iwata, H. “Changes in the concentrations of sotolon and furfural during the maturation of sake”, Journal of the Brewing Society of Japan, 99(5), 374–380, 2004. DOI 10.6013/jbrewsocjapan1988.99.374. Sotolon above its recognition threshold and furfural below it; odor unit of 61 in a 26-year-old sake. J-STAGE, free access
• Isogai, A., Utsunomiya, H., Kanda, R. and Iwata, H. “Changes in the Aroma Compounds of Sake during Aging”, Journal of Agricultural and Food Chemistry, 53(10), 4118–4123, 2005. DOI 10.1021/jf047933p. ACS Publications
• National Research Institute of Brewing (Japan). Identification of the component that causes hineka in sake, 2009 research results — DMTS-P1 as a precursor, and the doubling experiment. Underlying papers: Isogai et al., J. Agric. Food Chem., 57, 189–195 (2009) and 58, 7756–7761 (2010). PDF, English
• National Tax Agency (Japan). Standards for Manufacturing Methods and Quality Labelling of Sake (Notification No. 8, 22 November 1989, as amended) — the special designation categories, and 貯蔵年数 as an optional label item, rounded down, blends shown at the shortest figure. NTA, Japanese