The short answer: As a starting point we would use 100 g of coarsely ground coffee with 800 ml of cold water (1:8), steep it for 12 to 18 hours in the fridge, strain it and dilute it before drinking. That is our assessment, not a standard: the sources we read range from 1:20 to 1:4.3 and from 4 to 36 hours. Fuller and Rao (2017) found no further extraction of caffeine and chlorogenic acid after only 6 to 7 hours at 21 to 25 °C.
There is no binding definition of cold brew. Claassen et al. (2021) point this out and propose that only cold extracted coffee should be called that: hot extracted coffee should never be called cold brew. We follow that. Coffee that is brewed hot and chilled afterwards is covered in our guide Iced coffee at home.
The method, step by step
- Weigh the coffee and the water: 100 g of coffee, 800 ml of water. Hario, Fellow and AeroPress work in grams, and the ratio determines how strong the concentrate becomes. For its system, Toddy specifies filtered water; we did not look for a water hardness figure specifically for cold brew.
- Grind coarsely, roughly like sea salt (that is how Hario UK describes it). Why coarse is explained further below, because the reason is a different one from what is usually claimed.
- Put the grounds and cold water into a wide-mouthed jar and stir once so that nothing floats dry on top. That is our assessment; there is no manufacturer statement on it. With Toddy’s classic method (adding the coffee in layers) the opposite applies: do not stir, because a churned-up bed can clog the filter. With the paper filter bag variant, Toddy stirs gently.
- Cover and put in the fridge for 12 to 18 hours. That is close to the survey by Claassen et al. (2021): mostly in the fridge at about 8 °C, median steeping time 16 hours.
- Pour through a sieve or cloth, and if needed afterwards through a paper filter. The grounds have to come out completely, because the coffee keeps extracting as long as they stay in (Hario UK on the filter basket of the Mizudashi).
- Fill into a clean bottle, write the date on it, keep it cold.
Step 4 is our first fixed choice. 12 to 18 hours is one possible choice, not a documented value. If you start it in the evening, you have the concentrate in the morning.
Why cold water changes the calculation
Coffee releases its soluble substances into the water until an equilibrium is reached. How long that takes depends on the temperature. Batali et al. (2022) measured it at a ratio of 1:5 and a mean particle size of 972 micrometres: at 92 °C the equilibrium was reached after 1 to 2 hours, at 22 °C after 12 to 14 hours, at 4 °C after 24 to 36 hours. The hot brew also dissolved about 14% more (extraction yield 22.28% against 19.55% for the coldest).
For our batch this means: at 12 to 18 hours in the fridge and 1:5, we are still short of equilibrium. That is not a mistake, just a different strength. However, those times apply to 1:5 and this grind, not automatically to 1:8.
For caffeine and one chlorogenic acid (3-CGA), Fuller and Rao (2017) found equilibrium earlier. They used 35 g of coffee to 350 ml of water at 21 to 25 °C, took samples for up to 24 hours and saw no further significant rise in 3-CGA after about 400 minutes (6.7 hours); caffeine behaved similarly. Only these two substances were measured, neither taste nor total dissolved solids. An equilibrium of two substances does not mean the coffee tastes best after 6 to 7 hours.
The pilot study by Claassen et al. (2021) watched small laboratory runs by NMR. Without agitation, extraction rose quickly at first, then slowly: chlorogenic acid until about 120 minutes, and caffeine reached its plateau after about 140 minutes. The authors conclude that more than 7 hours is mostly not necessary at low turbulence. Ultrasound raised the measured compounds by 39% on average after one hour, constant shaking by 11%.
And the caffeine? With coarse grounds, it was clearly above that of a hot French press brew at the same ratio in cold brew: 1,230 against 970 mg/l for medium roast, 990 against 840 for dark roast. With medium grounds, the gap varied a great deal depending on the roast: 1,180 against 1,040 mg/l for medium roast, 1,080 against 1,060 mg/l for dark roast. The comparison was 24 hours cold against 6 minutes hot at 98 °C. Fuller and Rao attribute the difference to the limits of diffusion in the short hot brew.
The numbers: ratio, steeping time, state of the sources
First the ratio. The table shows what the sources state, sorted by the amount of coffee per litre of water (our own calculation, 1,000 divided by the water share).
| Source | Statement | Ratio | g coffee per litre |
|---|---|---|---|
| Claassen et al. 2021, survey of 125 people | 44% named this range, the most common | 1:20 to 1:10 | 50 to 100 |
| Lopane et al. 2024, laboratory experiment | 4 hours at 4 °C | 1:14 | 71 |
| Deutscher Kaffeeverband 2022, press release | 20 to 25 g in 250 ml for one cup | 1:12.5 to 1:10 | 80 to 100 |
| Cordoba et al. 2019 | 60 g coffee to 700 g water | 1:11.7 | 86 |
| Fuller and Rao 2017 | 35 g in 350 ml | 1:10 | 100 |
| Hario UK, Mizudashi | 60 g in 600 ml (concentrate 1:5, so 200) | 1:10 | 100 |
| Starbucks, according to our German review of the Espresso Roast | Pour-over fine, 20 g in 180 ml | 1:9 | 111 |
| Starbucks, same source | French press coarse, 22 g in 180 ml | 1:8.2 | 122 |
| Fellow, according to the review of the Opus | Dose 80 g at 1:8 | 1:8 | 125 |
| Our assessment | 100 g in 800 ml | 1:8 | 125 |
| Toddy, Brewing Guide 2022 | 340 g in 1.9 l | 1:5.6 | 179 |
| Batali et al. 2022 | 100 g in 500 g | 1:5 | 200 |
| AeroPress Inc., blog | Starting ratio | 1:5 | 200 |
| AeroPress Inc., overnight recipe | 30 g in 130 ml | 1:4.3 | 231 |
The Claassen survey is not representative. It ran through social networks and contacts of coffee roasters, and 71% of respondents brewed at home.
Our assessment: 1:8 sits in the middle between the association’s recipe and Toddy, matches Fellow’s dose figure and comes close to Starbucks. If you want to start milder, take 1:10 like Fuller and Rao, Hario and the Kaffeeverband. If you want a concentrate that can take a lot of water, go towards 1:7. Toddy and AeroPress start even stronger at 1:5.6 and 1:5, and the concentrate there is then no longer the same as ours.
Then the time. Here too the figures vary, and place and ratio play a part:
| Source | Place | Steeping time |
|---|---|---|
| Toddy 2022 | Room temperature | 8 to 24 hours |
| Hario UK | Fridge | 8 to 12 hours, 12 as the point for most coffees |
| AeroPress Inc., overnight recipe | Fridge | 8 to 12 hours |
| Deutscher Kaffeeverband 2022 | Room temperature or fridge | 4 to 12 hours or 12 to 24 hours |
| Deutscher Kaffeeverband 2016 | Room temperature | 12 hours |
| Claassen et al. 2021, survey | Mostly about 8 °C | Median 16 hours |
| Cordoba et al. 2019 | 20 °C | 14 or 22 hours, best sensory result at 14 |
Cordoba used two Colombian arabicas and three trained tasters. A rule of “the longer, the better” cannot be read from the table. Hario considers times of 16 hours and more rarely better. In Batali, the panel did not distinguish fridge (4 °C) and room temperature (22 °C) on bitter, sour and rubbery, only floral was higher in the coldest brew; the differences to the 92 °C brew were larger. For us, the fridge has a different reason, hygiene, more on that below.
Grind size: coarse yes, but for other reasons
Toddy prescribes coarsely ground coffee, Hario UK sea salt, the Kaffeeverband coarse. For the Opus, Fellow gives outer ring settings 10 to 11 for cold brew and 8 to 10 for the French press; the bars in the table are read off a graphic in the review. So cold brew is only partly coarser there. For Baratza, the instructions give the click settings 21 to 40 of the Encore ESP as filter to cold brew, without a number for cold brew. On the Virtuoso+, the suggestion table ends at the French press with setting 28, above which there are twelve click settings up to the coarsest setting 40. Both reviews would start above the French press.
The dissenting voice comes from Hario itself. In the instructions for the Skerton N, cold brew is given as fine to medium fine, the only line of the table that does not go beyond medium fine. The instructions do not say which device the line applies to. We suspect Hario’s cold extraction device, because in the table for the Mini Slim+ fine appears only for that device; that would be a different method from steeping in water. Hario points out that the grind can differ depending on the device.
What the studies say: Fuller and Rao (grind size not well controlled according to the authors, medium and coarse not sieved separately) found no significant difference between medium and coarse grounds for caffeine and 3-CGA, and explain it by diffusion running to completion in the long time. Cordoba et al. found higher values for TDS, extraction yield, phenols and titratable acidity with coarser grinding (701 to 900 against 501 to 700 micrometres) and longer time. The best sensory scores went to coarse with 14 hours.
None of the studies measures whether fine grounds over-extract or turn bitter in cold brew. Fuller and Rao do not measure bitterness at all, Cordoba has three tasters. The often heard line that fine powder turns bitter overnight is not backed by them. We would still grind coarse: Toddy, Hario UK, Fellow and the Kaffeeverband say coarse, and coarse grounds are easier to strain.
On the grinder: Stiftung Warentest (Germany’s consumer testing foundation) writes (issue 12/2019 with 15 electric grinders, plus test.de of 25 November 2024) that blade grinders often delivered less uniform particle sizes, while burr grinders, once set, give the same grind every day. Nobody there measured whether a burr grinder therefore gives a clearer cold brew. Comandante admits in its FAQ that its burr set too does not entirely avoid fines and coarse chunks, and Uman et al. (2016) found a bimodal particle distribution in ground coffee.
Diluting and mixing
Toddy recommends as a start 1 part concentrate to 2 parts water or milk, over ice or hot. It helps to know that Toddy’s concentrate at 1:5.6 is made much stronger than ours. Hario UK advises stretching the concentrate (there 1:5) with water or milk, but gives no figure, and AeroPress says “to taste”. How strong the drink in the glass becomes has not been measured so far.
We would start at 1:1 and add more water in small steps. The principle behind it: dilution and batch belong together. The stronger the concentrate, the more water it needs. If you start at 1:10, you dilute less than if you work with 1:5.
For a cold brew latte we would take 1 part concentrate to 2 to 3 parts milk, an assessment without a source. In 2016 the Kaffeeverband showed recipes with cold brew concentrate, tonic water and cherry syrup (Cherry Bomb), but gave no amounts.
Acidity, bitterness, sweetness: what was measured
“Cold brew has less acid” is true, but only for one of the measures and for light roasts under these conditions. Rao and Fuller (2018) used 35 g of light roasted coffee to 350 ml, cold for 7 hours at 21 to 25 °C, hot for 6 minutes in the French press. The pH was comparable (cold 4.96 to 5.13, hot 4.85 to 5.10). The titratable acidity was lower when cold: 4.75 to 5.88 against 6.40 to 7.85 ml of 0.10 molar sodium hydroxide per 40 ml. The antioxidant activity was also lower. The authors note that pH reflects acidity differences poorly.
On perception there is the study by Batali et al. (2022). A trained panel tasted the same coffees at the same strength of 2% TDS and described the hot brewed, chilled coffee as more bitter, more sour and more rubbery, and the cold one as more floral. That holds across all roasts and origins, but the effects of brewing temperature were clearly smaller than those of roast level and origin. The coffee you choose shapes the cup more than the method, as the SCA article on the study sums it up. 27 samples were examined from three coffees, three roast levels and three temperatures. The study was funded by the Coffee Science Foundation, with support from Toddy LLC, a maker of cold brew devices; Toddy also supplied the devices.
Sweetness is not reported by the study. Cordoba et al. describe the cold brews of their three tasters as sweet with notes of malt, almond, cocoa and blossom, but rated them lower than the hot ones for aroma, taste, aftertaste, acidity and body. In a triangle comparison by Claassen et al., 20 of 25 assessors recognised the difference between hot and cold brew, 17 preferred the cold brew, 6 the hot one, 2 had no preference. In the ranking there was no significant difference. “Milder” and “sweeter” are therefore not facts, but descriptions from small panels.
Shelf life and hygiene
Here the sources vary the most. For its system and recipe, Toddy gives up to 2 weeks, the Kaffeeverband also named two weeks in 2016, both without independent checking. Hario UK writes about one week in a closed container. The Wisconsin Department of Agriculture (DATCP, a US state authority, fact sheet for food service businesses, as of December 2023) classes cold brew without a heating step as a TCS food, that is, one that needs refrigeration. Without a HACCP plan, a maximum of 41 degrees Fahrenheit (5 °C) is allowed and date marking of no more than seven days. The fact sheet names temperature and time abuse, reduced oxygen packaging such as kegs and nitrogen, and poor hygiene as risks. With correct preparation, storage and serving, the risks of illness are low, it says.
The most cautious are Claassen et al. (2021), who draw their conclusions from pilot data from cafés and not from a storage study. They examined 23 cold brew samples from cafés in southern Germany, transported at 8 °C in summer 2020. Two samples (9%) had an elevated microbial load. In one they found presumed Bacillus cereus, and it had been stored for five days; the other had clearly elevated lactic acid bacteria and yeasts. The authors recommend not keeping cold brew longer than one day and discarding it if the smell or taste changes.
We found no statement on cold brew in the household from the BfR, the Verbraucherzentrale (German consumer advice centre) or the Umweltbundesamt (German Environment Agency) in our research on 14 October 2026. For mixtures with milk we found no source at all; we would only mix them fresh and not keep them in stock. Our assessment, not a limit value, for concentrate without milk: keep it cold throughout, in a clean container, for at most about a week, and smell and taste it before diluting. If you take Claassen’s one day seriously, make smaller batches.
Cold brew or iced coffee
The difference lies in the water when you start the brew, not in the glass at the end. With cold brew it is cold from the start; the beans were of course roasted. Iced coffee is inconsistent in practice: in 2016 the Kaffeeverband showed recipes with filter coffee, espresso and instant coffee, and Vietnamese cà phê sữa đá is brewed hot and dripped onto ice. There is therefore no temperature that defines iced coffee. The 92 to 96 °C that are occasionally named for iced coffee are a criterion for filter machines (according to the manufacturer, Melitta names it for the Golden Bean award of the ECBC, SCA 310 names 90 to 96 °C for machine evaluations, details in the FAQ) and not a feature of the drink.
Troubleshooting
| Symptom | Possible cause | What you change |
|---|---|---|
| Concentrate tastes thin | The batch is weaker or you dilute more than needed. The strength follows from ratio and dilution. | Next batch towards 1:7, or add less water |
| Concentrate is cloudy, sediment at the bottom | Fines in the grounds get through the sieve | Use a paper filter (Toddy works with a paper filter bag), grind coarser |
| Sieve or filter clogs | Churned-up bed of grounds, as Toddy warns for its classic method | With this method do not stir, add the grounds in layers |
| Tastes bitter | In Batali, roast level and origin had a stronger effect than brewing temperature. | Try a different bean or a lighter roast, shorten the steeping time (Cordoba: shortest time, best sensory result) |
| Smells or tastes different from fresh | Possible spoilage. Claassen et al. advise discarding if it changes. | Pour it away, make the next batch smaller |
| Cold brew comes out warm from the bean-to-cup machine | According to the troubleshooting section in our German review of the La Specialista Touch, a hot drink had been made shortly before | Empty the tank before the cold brew, rinse and refill with fresh water |
Which device, which bean
We do not have a review of the Hario Mizudashi, Toddy or any other dedicated cold brew vessel. What exists are devices that can also do cold brew.
Amazon explicitly lists the Amazon Basics French Press with cold brew as a possible use of the 1 litre jug, and the plunger unit separates the grounds. Our German review advises measuring the height of the fridge shelf beforehand. For the OZAVO Thermo, the box says: room temperature water, 6 to 12 hours at room temperature or in the fridge, then press. The box does not give a coffee amount; here our dosing by weight applies. For the Bodum Brazil and the Groenenberg stainless steel thermal jug, the sources of our reviews do not answer the question, and Bodum doses in spoonfuls. We do not recommend either of them for this.
Our German review classes the AeroPress Original as technically possible, but recommends clearly more time than 60 seconds. For the AeroPress Go, the manufacturer’s text mentions cold brewed coffee; the device delivers the concentrate, it does not come with milk frothing.
The De’Longhi La Specialista Touch has a cold extraction function. According to De’Longhi it delivers cold brew in under 5 minutes per single portion, adjustable from 90 to 220 ml. Compared with our 12 hours of steeping, that is a fraction: 5 minutes against 12 hours makes 144 times, against 8 hours still 96 times (our calculation). If you drink by the litre, stay with the fridge batch from this guide. According to our reviews, cold brew programmes are also found on the Ninja Luxe Café Pro, the Siemens EQ600 and the Sage Barista Touch; what they do is in the FAQ above.
On choosing: hot filter coffee follows different rules, which are in the guide Pour-Over and Filter Coffee. If you take figures from elsewhere, treat them as starting values and adjust your concentrate and the dilution to your own taste.