Koji Propagation — Aspergillus oryzae on Grain
One of 3 entries · The Noma Guide to Fermentation — Redzepi/Zilber 2018
Aspergillus oryzae cultivation on steamed grain dates back over a thousand years in Japan, China, and Korea, forming the enzymatic backbone of miso, sake, soy sauce, mirin, and amazake. The technique was systematised in Japanese breweries (kura) and documented in detail by Shizuo Tsuji as foundational to the logic of Japanese cuisine.
Koji is a mold, not a seasoning. Propagating it means giving Aspergillus oryzae the right temperature, humidity, and oxygen to colonise steamed grain — typically rice, barley, or wheat — and produce a dense network of enzymes: amylases that break starches into fermentable sugars, proteases that cleave proteins into amino acids and glutamates, and lipases that work on fats. What you get after 40–48 hours is a grain that smells like chestnuts and warm mushrooms, tastes faintly sweet and deeply savoury, and carries enough enzymatic activity to transform whatever you bury in it next — meat, fish, vegetables, dairy. The cook's job is environmental. Soak and steam the grain until it's fully cooked but not wet on the surface — excess surface moisture drowns the mold before it takes hold. Inoculate with tane-koji (spore powder) once the grain has cooled to around 30–35°C. Spread evenly. Then manage a 40–48 hour incubation at 28–32°C with 70–85% relative humidity, aerating the mass every 12 hours or so by hand-mixing (called te-ire in traditional brewing), which disperses heat generated by the mold's own metabolism and prevents hot spots that kill the culture or push it into sporulation too early. By hour 20–24, mycelium should be visible as white filaments binding grains together. By hour 40, the mass should hold together when pressed, smell intensely of roasted chestnut and fermented grain, and feel warm and slightly dry on the surface. Sporulation — a green-grey colour — signals you've gone too long; the mold has shifted from enzyme production into reproductive mode, and enzymatic yield drops sharply. In a modern kitchen context this means owning a dedicated incubation chamber with a temperature controller, a humidity source (ultrasonic humidifier or wet towels with a probe), and a perforated tray system so airflow stays consistent around the entire mass. Koji made this way is a working ingredient — a fermentation engine — not a flavouring in the conventional sense.
- Meju (Korean): fermented soybean blocks inoculated with Aspergillus, Rhizopus, and wild Mucor species — precursor to doenjang and ganjang; shares the principle of mold-driven protease and amylase activity but uses wild rather than controlled inoculation.
- Qu (Chinese): compressed fermentation starter for baijiu and huangjiu production using Aspergillus, Rhizopus, and yeast on wheat or barley; enzymatic logic parallel to koji but with deliberate multi-organism complexity.
- Tempeh starter (Indonesian): Rhizopus oligosporus on soybeans — similarly a mold-propagated grain product producing protease activity and binding mycelium, though flavour profile and fermentation timescale differ significantly from Aspergillus oryzae.
- European bread malt: barley germinated to produce amylase activity for sugar conversion in brewing and baking — analogous enzymatic function (starch-to-sugar conversion) achieved through germination rather than mold propagation.
Aspergillus oryzae secretes extracellular proteases — notably acid proteases and neutral proteases — that cleave protein chains into free amino acids including glutamic acid, the primary driver of umami perception. Simultaneously, alpha- and glucoamylases hydrolyse starch into glucose and maltose, producing sweetness and fermentable substrate for downstream microorganisms. The chestnut-mushroom aroma is largely attributable to volatile compounds including 2-acetyl-1-pyrroline and various furanones produced during the mold's metabolism. This is why finished koji tastes simultaneously sweet, savoury, and faintly floral — three sensory dimensions arising from a single microbial process acting on two different macronutrient classes in the grain.
Steam grain until cooked through but surface-dry; standing moisture on the kernel prevents mycelial penetration and promotes bacterial contamination. Inoculate at 30–35°C — below 38°C kills the mold slowly; above 40°C kills it fast; below 25°C stalls germination entirely. Maintain 70–85% relative humidity throughout; drop below 65% and the mold desiccates; above 90% and unwanted bacteria compete. Perform te-ire — hand-turning the mass — every 10–14 hours to redistribute metabolic heat and prevent runaway temperature spikes in the core. Pull the koji at peak mycelial coverage before visible sporulation; green or grey colouration signals enzyme production has peaked and is declining. Use food-grade tane-koji from a verified supplier with known spore count; wild inoculation is inconsistent and unsuitable for controlled production.
{"Line perforated hotel pan inserts with dampened cotton cloth rather than plastic wrap — cloth regulates surface moisture passively, wicking excess while maintaining humidity, and gives the mycelium something to grip during early colonisation.","Log grain core temperature every 4 hours with a probe thermometer; if core temperature exceeds 38°C before the 36-hour mark, pull the tray immediately, turn and spread thin, and allow a 30-minute rest before returning it — this preserves enzymatic yield when incubation rooms run warm.","Freeze finished koji in sealed bags within 2 hours of pulling; enzymatic activity continues at refrigerator temperatures and the koji will over-ferment any product it contacts; freezing halts activity while preserving the enzyme profile intact for kitchen use.","Test your finished koji's protease activity empirically: press a small amount against a raw chicken breast and hold at 55°C for 30 minutes — the surface should show measurable texture breakdown and pronounced savoury aroma; weak or absent breakdown signals poor protease yield from suboptimal propagation."}
Under-drying the grain surface after steaming: free surface moisture creates anaerobic pockets that favour lactic acid bacteria and suppresses Aspergillus, producing sour, off-smelling koji with low enzymatic activity. Skipping or delaying te-ire: the mold generates significant metabolic heat as it grows; an unturned mass can hit 42–45°C in the core within 18 hours, killing the culture at the centre while the outer layers appear normal. Pulling too late: koji left past 48–52 hours in warm conditions begins sporulating visibly; protease and amylase activity both decline as the organism shifts biological priority toward reproduction, reducing the functional value of the finished koji. Using an incubation environment that lacks airflow: stagnant humid air at the tray surface creates wet patches, encourages Mucor and other competitive molds, and produces an uneven, patchy colonisation that is difficult to use reliably.
The Noma Guide to Fermentation — Redzepi/Zilber 2018
Kitchen membership opens the full Library.
Grain soaked and steamed to precise gelatinisation with zero surface sheen; inoculated at 32°C; incubated… Good surface drying of grain; inoculation at 30–35°C; temperature-controlled incubation chamber at 28–32°C with active…
visual: At 40–44 hours, mycelium should appear as a continuous white mat binding grains together into a cohesive cake that…
Where the dish lives or dies: grain surface moisture at the point of inoculation — too wet and the spores germinate in free water rather…
Common Questions
Why does Koji Propagation — Aspergillus oryzae on Grain taste the way it does?
Aspergillus oryzae secretes extracellular proteases — notably acid proteases and neutral proteases — that cleave protein chains into free amino acids including glutamic acid, the primary driver of umami perception. Simultaneously, alpha- and glucoamylases hydrolyse starch into glucose and maltose, producing sweetness and fermentable substrate for downstream microorganisms. The chestnut-mushroom aroma is largely attributable to volatile compounds including 2-acetyl-1-pyrroline and various furanones produced during the mold's metabolism. This is why finished koji tastes simultaneously sweet, savoury, and faintly floral — three sensory dimensions arising from a single microbial process acting on two different macronutrient classes in the grain.
What are common mistakes when making Koji Propagation — Aspergillus oryzae on Grain?
Under-drying the grain surface after steaming: free surface moisture creates anaerobic pockets that favour lactic acid bacteria and suppresses Aspergillus, producing sour, off-smelling koji with low enzymatic activity. Skipping or delaying te-ire: the mold generates significant metabolic heat as it grows; an unturned mass can hit 42–45°C in the core within 18 hours, killing the culture at the centre while the outer layers appear normal. Pulling too late: koji left past 48–52 hours in warm conditions begins sporulating visibly; protease and amylase activity both decline as the organism shifts biological priority toward reproduction, reducing the functional value of the finished koji. Using an incubation environment that lacks airflow: stagnant humid air at the tray surface creates wet patches, encourages Mucor and other competitive molds, and produces an uneven, patchy colonisation that is difficult to use reliably.
What dishes are similar to Koji Propagation — Aspergillus oryzae on Grain?
Meju (Korean): fermented soybean blocks inoculated with Aspergillus, Rhizopus, and wild Mucor species — precursor to doenjang and ganjang; shares the principle of mold-driven protease and amylase activity but uses wild rather than controlled inoculation., Qu (Chinese): compressed fermentation starter for baijiu and huangjiu production using Aspergillus, Rhizopus, and yeast on wheat or barley; enzymatic logic parallel to koji but with deliberate multi-organism complexity., Tempeh starter (Indonesian): Rhizopus oligosporus on soybeans — similarly a mold-propagated grain product producing protease activity and binding mycelium, though flavour profile and fermentation timescale differ significantly from Aspergillus oryzae.