Scope. This page provides a controlled reader-run protocol and does not claim site-run measurements. For cooking, cooling, storage and reheating, follow the current official guidance linked on the page.
Somewhere in the twentieth century a number attached itself to caramelised onions and never came off. It gets restated because it has always been restated, and because the person restating it is usually describing a real thing they cooked, in a pan whose diameter they did not mention, at a quantity they did not mention, to a definition of "done" they did not state. Every one of those omissions moves the answer by more than the difference people argue about.
The result is a cook standing over a pan an hour in, onions still amber rather than mahogany, concluding that they have done something wrong. They have not. They are being measured against a number nobody re-timed.
This page provides a timing protocol rather than a universal duration. Record pan geometry, onion mass, heat setting, stirring intervals and the time at which the stated colour endpoint is reached; report the observed time only with those conditions.
Define "done" before you time anything
A timing claim is meaningless without a stated endpoint, and this is the whole reason the published figures disagree. There are at least four endpoints in common use, and they are hours apart at the far end.
| Endpoint | What it looks like | What it tastes of |
|---|---|---|
| Softened | Translucent, slumped, no colour | Onion, mild, still sharp at the edges |
| Golden | Pale gold, still holding shape | Sweetening, recognisably onion |
| Browned | Even mid-brown, collapsed, volume much reduced | Sweet, some depth, some residual bite |
| Caramelised | Deep mahogany, jam-like, no distinct pieces remaining | Sweet throughout, savoury, no raw note anywhere |
The control run drives to the fourth state, defined for the log as: no piece retaining its original structure, uniform deep colour with no pale fragments, and no raw allium note detectable on tasting by two people independently. That definition is written into the log before the pan goes on so that the clock cannot be stopped by wishful thinking.
Most published timings describe the second or third state. The disagreement is mostly definitional rather than factual, which is worth knowing before anyone accuses anyone of lying. The lie in the title is not a person. It is a convention that survives because the endpoint is never specified.
The control run
The variables that move this result most are the ones almost always omitted, so they are the ones to record first.
- Quantity. Stated by mass, not by "two large onions." Volume in a pan drives the whole timing.
- Pan diameter and material. A wide heavy pan and a narrow one are performing different processes at the same hob setting, because the wide one evaporates faster.
- Depth of onion in the pan. The single best predictor of how long the water takes to leave, and derivable from the two figures above.
- Cut. Thickness and direction, both photographed.
- Fat. Type and mass.
- Medium temperature, logged continuously. Pan-surface temperature at intervals, not a hob dial setting.
- Ambient kitchen temperature, since evaporation rate is not independent of it.
- Stirring frequency, which is a variable even though it is usually written as an instinct.
Readings and photographs are taken at fixed intervals through the run rather than at the end, because the shape of the curve is the informative part. The full protocol is on how we test.
For the control, record the measured medium temperature, total time to the stated endpoint, starting and finishing mass, calculated water loss, and replicate count. Keep all of those values tied to the same equipment and run.
The stages, and where most people stop
Onions do not brown steadily. They pass through phases that behave differently, which is why a single average rate is misleading even when the total is right.
Phase one: water leaves. Onions are mostly water. Until enough of it has gone, the surface cannot exceed the boiling point, so browning cannot begin regardless of the hob setting. Adding salt early accelerates this by drawing water out, at the cost of a wetter pan for a while. This phase is pure evaporation, and its length is set by pan area and quantity rather than by heat. The same ceiling is explained in why vegetables steam instead of browning.
Phase two: colour begins. Once the surfaces are dry enough, browning reactions start and colour appears at the edges. This phase feels fast and encouraging, and it is where a lot of people declare victory.
Phase three: the long middle. Colour deepens slowly and evenly, volume keeps dropping, and very little appears to happen from one glance to the next. This is where most of the clock goes, where the pan needs the most attention against sticking, and where nearly everyone stops early.
Phase four: jam. Structure gone, colour uniform, the mass now sticking readily and needing deglazing with a splash of water to lift the fond back in. The transition into this state is the least documented part of the process and the part most worth photographing.
We give no share of the total for each phase. Note the time at which each one begins in your own pan and you will see where your clock goes.
Shortcut one: bicarbonate of soda
The alkaline batch is included as an experiment, not a recommendation. Record whether it colours faster than the control under the same measured conditions and what happens to texture and flavour; this article makes no claim about how much faster it is.
The costs reported by cooks who use it are texture and flavour rather than speed. Alkalinity also breaks down pectin, so onions treated this way tend toward mush faster, and past a small dose there is a soapy or metallic note. Both of those are judgements, which means they get tested blind rather than declared.
How to test it. Same mass, same pan, same temperature, same endpoint definition, with a stated dose. Two panels: one timing the run to the endpoint, one tasting the finished product against the control without knowing which is which, scoring sweetness, depth, off-notes and texture separately.
What to record: the dose; the time to the endpoint beside the control's; the blind scores for sweetness, depth, off-notes and texture; and a photograph of both pans at the endpoint.
If it measurably helps in your pan, use it. A shortcut that saves real time at an acceptable cost is a useful thing to own, and dismissing it on principle would be its own kind of dishonesty.
Shortcut two: added sugar
Adding sugar puts caramelisation reactants directly into the pan, so colour arrives sooner. The question is whether the result tastes like caramelised onions or like onions with sugar in them, and that is a tasting question rather than a timing one.
The suspicion worth testing is that sugar produces the appearance of the endpoint before the onions have done the structural work that gives the real thing its texture and savoury depth. If that is right, the shortcut is not faster so much as it is a different dish that photographs the same. If it is wrong, it is a genuinely useful weeknight tool.
How to test it. Identical protocol, stated dose, blind tasting against the control, with texture scored separately from flavour so the two cannot be conflated.
What to record: the dose; the time to the colour endpoint beside the control's; whether the structure had broken down as far as the control's at that point; and the blind scores for flavour and texture, kept separate.
Shortcut three: higher heat with added liquid
The most common experienced-cook approach: run the pan hotter than the traditional advice, and add small amounts of water whenever it threatens to catch, deglazing the fond back into the onions each time.
The logic is sound. The fond on the pan base is browned material, and lifting it back in distributes colour and flavour without letting it burn. The added water also holds the pan surface temperature down briefly at each addition, so the run is effectively a series of short high-heat pushes separated by resets.
What has to be measured is whether the finished result matches the control or merely resembles it, and how much attention the method costs. A technique that saves twenty minutes but requires you to stand at the pan the entire time is not a saving for most weeknights.
How to test it. Same endpoint, temperature logged through every addition so the actual thermal history is visible, plus a count of interventions.
What to record: the time to the endpoint beside the control's; the pan temperature before and after each water addition; the number of additions and how long you had to stand at the pan; and the blind comparison against the control.
The honest number, and how to live with it
The real fix is not a shortcut. It is scheduling.
Caramelised onions keep well, they reheat without loss, and they are one of the few preparations that improve a dish out of all proportion to the effort of adding them. Making them on a weeknight, from raw, in the time between arriving home and eating, is the actual mistake. Nothing in the chemistry cooperates with that plan.
- Cook a large batch at once, in the widest heavy pan you own, on a day when the clock is not the constraint. The marginal time cost of tripling the quantity is far less than tripling the time, though it is not zero; record how long your larger batch actually takes.
- Portion and refrigerate or freeze. They freeze well in flat sheets that snap into portions.
- Under-cook slightly if you are freezing, since reheating in a pan will take them a little further.
- Keep the fond. Deglaze the pan at the end and stir the liquid back in rather than washing it away.
Storage times and conditions are a food-safety question, not a texture question, and we will not estimate them. For cooling and storing cooked vegetables, follow the USDA FSIS leftovers guidance.
In Plain English
What it means. Caramelised onions take considerably longer than the number most recipes print, mostly because those recipes are describing a paler endpoint and are silent about pan size and quantity. If yours are not done at the promised time, the promise was the problem.
Who it helps. Anyone who has stood over a pan feeling incompetent, and anyone who wants to know whether the bicarbonate trick is real before trying it on a dinner they care about.
When to use it. Any time a recipe gives you a browning time without giving you a quantity and a pan diameter, treat the time as an estimate for a pan that is not yours.
What to do next. Decide which endpoint you actually want, since golden onions are a perfectly good ingredient and take a fraction of the time. If you want the deep version, make it in advance and in bulk. The broader framing of when to sweat and when to brown is in sweating versus browning, and the rest of this section is at everything in this section.
Keep your timed run
Keep the control, the three shortcuts, the blind-tasting scores, and failed runs together as one evidence record. This page provides the complete comparison protocol and does not claim a site-owned timing result.