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.
A whole fish is the least forgiving thing most people ever cook, and it is usually cooked once. There is no practice run, because the practice run costs as much as the real one. You get a single attempt, and the standard instruction available to you is minutes per kilogram, which is a formula that assumes an oven whose real temperature you have never checked, a fish of a shape it does not know, and a starting temperature nobody asked about.
This post replaces the timer with a number. You cook until a specific reading appears at a specific place in the fish, you account for what happens after it leaves the heat, and you accept a longer cook in exchange for a much wider margin for error. Cooking gently is not a philosophical preference here. It is the only way to make the window wide enough to hit.
Species-specific safety guidance, including the cases where a lower target is not appropriate, is collected and cited on the guide. Read the doneness windows by species before you commit to a target for the fish in front of you. Nothing on this page overrides it.
Your Heat Spec
This page gives no Heat Spec values of its own: the target belongs to the species, and every other number depends on your fish and your oven. Keep one line per cook in a notebook with these fields: species and weight; medium temperature as measured by a probe, not the dial; the fish's starting internal temperature; your target, taken from the species guide; your pull temperature; the highest reading after the rest; and the time at each reading.
Photographs. Photograph your own fish at the pull and again after serving, and keep the pictures with the notes. We do not generate food imagery, and a photograph of a failure teaches more than one of a success.
Buying, And Asking For The Right Preparation
The fish is the variable you have the most control over and the least information about, so spend the attention here.
Judge the fish physically rather than by label. The checks that actually correlate with freshness, and the ones that are folklore, are set out in how to judge a piece of fish. Do that first, at the counter, before you get attached to a plan.
Then ask for the preparation you want, specifically. A good counter will do all of this and it costs nothing.
- Scaled and gutted. Non-negotiable, and vastly easier for them than for you.
- Gills removed. Frequently skipped, and they carry a bitterness into the head end.
- Fins trimmed. Makes handling and serving less hazardous.
- Left whole otherwise. The head and frame are structure. They hold the fish together during cooking and they hold flavour.
Ask how it was stored and when it came in. You are not testing them; you are collecting the two facts that most affect what you are about to do.
Size is a real decision. A fish that overhangs your largest tray will cook unevenly no matter what you do, because part of it will sit against the oven wall. Buy to your equipment. Two smaller fish are easier to cook well than one large one, and they do not need proportionally longer.
Prep: Scoring, Cavity, Seasoning Timing
Scoring. Two or three cuts through the skin into the thickest part of each side, angled, stopping short of the bone. The purpose is to let heat into the thickest region so that the thick and thin parts of the fish converge rather than diverge. It also gives you an honest visual check on the flesh later. Do not over-score a small fish; you are managing evenness, not decorating.
Cavity. Aromatics in the cavity do less than people hope, because the flesh is dense and the contact time is short. What they do reliably is scent the steam inside the fish and perfume the frame. Keep it loose. A packed cavity insulates the interior and slows the whole cook from the inside, which is the opposite of what you want.
Seasoning timing. Salt on the skin before cooking changes surface moisture over time. How long that takes on a whole fish, and whether it measurably changes the interior, is a variable for the reader-run protocol. Measure and record it on your own fish and equipment; this page supplies no optimum interval.
Temperature at the start. Take the fish out of the fridge before you cook it. A fridge-cold fish and a room-temperature one take different times to arrive at the same reading, which is one more reason the minutes-per-kilo formula fails. Rest it at room temperature for a stated interval and record it, because it is part of the experiment.
Where To Probe A Whole Fish, And Where Not To
This is the single point of failure of the whole method. A correct target read in the wrong place is worse than no reading at all, because it gives you confidence.
Where to probe. Into the thickest part of the flesh, behind the head and above the spine, entering from the top of the fish at a shallow angle so the tip travels along the fillet rather than across it. You are aiming for the geometric centre of the thickest muscle. Advance slowly and watch the reading: it will fall as you approach the centre and rise again as you pass through toward the far side. The lowest reading you can find is the one that matters.
Where not to probe.
- Against the bone. Bone conducts differently from flesh and will read high. This is the commonest error, and it makes an undercooked fish look done.
- Into the cavity. You will read hot air.
- Through a score line. The flesh there is exposed and hotter than the interior.
- The tail end. It is thinner and further along. It will always read higher, and it is not the part that decides the fish.
Instrument. A thin-tipped, fast probe is worth more than an accurate slow one here, because you are hunting for a minimum by moving the tip and you need the reading to keep up. How to check a probe's response time and accuracy against ice and steam is set out in how to test six thermometers.
Calibration. An uncalibrated probe reporting a number to one decimal place is a confident guess. Check it before you cook something expensive.
The Cook, With Readings At Each Stage
Whether you use a low oven or a gentle poach, the shape of the method is the same: a medium held well below the boil, a slow approach to target, and a decision made on a reading.
- Bring the medium to temperature and confirm it. Oven dials are frequently wrong, and by more than people expect. Read the actual cavity temperature with a probe, or read the poaching liquid directly. If your oven runs off target, adjust the dial until the measured temperature is right and ignore the number printed on the front.
- Fish in, positioned flat. If it is curling, it will cook unevenly. A support under the belly or a shallow bed of aromatics keeps it level.
- First reading, early. Take a reading well before you expect anything, so you know the shape of the curve rather than only its end. Note the time and the value.
- Second reading, at roughly two thirds. From two readings you can estimate the rate of climb, and from the rate you can predict when to be standing at the oven. This is the substitute for a timer, and it adapts to your fish and your equipment automatically.
- Then stay close. The last few degrees arrive faster than the first ones, because the temperature gradient across the flesh has narrowed. This is where fish is lost.
- Pull on the number, not on the clock. Write down the reading and the time at each stage; that record is what makes your next whole fish easier than this one.
The gentler the medium, the slower the climb and the wider the window in time terms. That is the entire argument for cooking a whole fish at a low temperature. The same doneness window in degrees can be a couple of minutes wide in a hot oven and considerably wider in a cool one.
Carryover, And When To Pull
A whole fish continues to cook after it leaves the heat. Heat stored in the outer flesh, in the frame and in the head keeps moving inward, and the reading at the centre keeps climbing for a while after the fish is on the counter.
Two things follow.
You pull below your target, not at it. The amount you pull below is the carryover figure, and it is specific to the size of the fish and the temperature of the medium, so we give no single number. Measure yours by leaving the probe in after the pull and noting the highest reading. A large fish out of a low oven behaves differently from a small one out of a hot one, and a single published carryover number that ignores both is not useful.
And the rest is part of the cook, not a pause after it. Rest it somewhere warm and uncovered, or loosely tented. Wrapping it tightly traps steam against the skin and softens it, which undoes the only textural benefit the skin had.
If you are going to make one change to how you cook fish, make it this one. Most overcooked fish was pulled at the target rather than below it.
Serving, And What To Do With The Frame
Serve from the top fillet first: run a knife down the spine and lift the two sections off, then lift the frame away from the bottom fillet in one piece rather than turning the fish over. Turning it tears the flesh, and there is no advantage to it.
The frame, the head and the collar are the most rewarding parts of the transaction and the most often binned. The collar carries the fattiest, most forgiving flesh on the fish, and the cheeks are a small, dense muscle worth digging out. Whatever is left after that is a stock ingredient, and fish frames want a short, gentle extraction rather than a long one. Everything in our fish section treats the frame as part of what you paid for.
In Plain English
What it means. A whole fish is cooked to a reading in a specific place, not to a time. The reading is taken at the thickest part above the spine, avoiding bone, and you remove the fish below your target because it keeps cooking afterwards.
Who it helps. Anyone who has bought one expensive fish for an occasion and does not want to find out at the table that the centre is raw or the flesh has gone chalky.
When to use it. Any whole round fish you can fit flat in your largest tray or pan, cooked in a low oven or a gentle poach.
What to do next. Check the doneness window for your species on the guide. Calibrate your probe. Salt the fish and let it come up from fridge cold. Take an early reading, take a second, then pull below your target and let carryover finish the job.
Next Steps
Before you buy, read how to judge a piece of fish. Before you set a target, read the doneness windows by species, where the safety guidance is cited to named authorities.
The instrument matters more than anything else in this article, and an uncalibrated probe is the reason most of these methods fail. Check your probe first with the calibration method; it takes two minutes.