Freeze to a -35°C core, or hold the core at -20°C for 24 hours, and never freeze a fish that sat warm.
How much anisakis is actually in a saltwater fish?
Japan recorded an average of 19,737 anisakiasis cases a year across 2018 and 2019, calculated from health insurance claims rather than from outbreak reports. Molecular work on the larvae found that 88.4% of those patients carried Anisakis simplex sensu stricto, a single species out of the group usually lumped together as whale worms. That is the largest national figure published for this parasite, and it comes from a country with both a raw-fish culture and a claims system that sees the cases.
The US regulation takes a narrower position. Under 21 CFR 123.6(b)(vii), parasites become a hazard requiring control only where the processor knows or has reason to know that the fish will be eaten without a process sufficient to kill them. Read that again. The rule does not treat a fish as dangerous. It treats raw consumption as the thing that switches the parasite from background to hazard.
That distinction is the whole decision. A mackerel you are going to grill has one handling standard. The same mackerel, headed for a plate of sashimi, has a stricter one, and the difference is created by you.
The larvae that matter sit in the muscle as well as the gut lining, which is why the Japanese freezing study embedded larvae inside mackerel fillets instead of testing them loose in a petri dish. Loose larvae die fast. Larvae inside flesh are protected by the tissue around them, and that enclosed version is what you are actually dealing with.
Reported incidence tracks reporting infrastructure as much as it tracks fish. Japan's claims-based count catches cases that other countries log as sporadic food poisoning or never log at all, so a low national number is not evidence of a low parasite load in the water. It is evidence of a shorter paper trail.
Why the freezer dial is the wrong number to read
A 2025 study from the Tokyo Metropolitan Institute of Public Health ran two freezers against each other: an air-blast rapid freezer and a natural-convection unit set at -20°C. Loose larvae died after 8 minutes in the rapid freezer and after more than 2 hours at -20°C. Inside mackerel, at the moment the fish core reached -20°C, a few larvae were still alive. When the core reached -35°C, all of them were dead.
The timing is the part people get wrong. In the same experiment, the time for the core to reach -20°C had a standard deviation under 15 minutes with rapid freezing and 171 minutes with conventional freezing. A domestic freezer in still air sits at or below the slow end of that range. Closing the lid starts the clock, but the core is still warm for hours.
So the same nominal 24 hours produces different results depending on when the core got cold. The study's conclusion is precise: all larvae died after 24 hours at a core of -20°C, but a few survived 24 hours from the start of freezing at -20°C, because the core had not been there for the full period.
Measure the thickest part of the fish rather than the air around it. If you have no probe, treat the clock as unreliable and add margin. A home freezer at -18°C is 2°C warmer than the study's endpoint, which is the wrong direction, so the hold gets longer, not shorter. A week at -18°C is a working rule for fillets that are already firm and thin.
Geometry beats equipment. A 2 cm fillet and a 6 cm loin in the same freezer are not the same experiment, because the loin has a core that lags for hours while the fillet is almost all surface. That is why the study reports a number for the core and not for the air, and why butterflying a loin before freezing is a technique rather than a habit. What a home freezer can and cannot do is covered in home freezer limits.
Where does histamine come from, and why can't the freezer fix it?
Histamine is not a parasite problem and freezing does not solve it. Bacteria on the fish convert free histidine into histamine, and they do it fastest when the fish is warm. Freezing stops the bacteria. What they already made stays in the tissue, and the toxins responsible are heat-stable, so cooking a contaminated fillet does not reduce the risk.
Singapore's food authority sets a limit of 100 ppm in fish. In a 2022 case report, two people who shared a tuna dinner were tested after symptoms that began within an hour, and the sample from the patient's home measured 6,556.68 ppm, roughly 65 times the limit. The samples taken from the distributor were within normal limits, which placed the cold-chain breach at the consumer's end of the chain.
The implicated species are not a short list. The name comes from Scombridae, the family that includes mackerel and tuna, but mahi-mahi, salmon and sardine have also caused cases. What they share is free histidine in the muscle and a fast route to bacterial spoilage once the fish stops being cold.
Scale is easy to underestimate because the illness mimics an allergy: rash, flushing, headache, nausea, and in bad cases a drop in blood pressure. A 2013 US estimate put the count above 35,000 cases with 162 hospital admissions between 2000 and 2009. Those cases were created by handling rather than by the fish.
Under-reporting explains why the numbers look inconsistent between countries. The same review calls scombroid an under-reported illness, and the reason is structural: symptoms fade within a day, most people never see a doctor, and the fish is usually eaten before anyone thinks to keep a sample. The Singapore case only produced a measurement because the patients kept the tuna and the authority tested it.
The fish that need the fastest chill
Rank by histidine first and by parasite load second. Tuna, mackerel, bonito, mahi-mahi, sardine and anchovy go into the ice the moment they are boated, because their muscle is the raw material for histamine. A yellowfin that thrashes on deck for ten minutes in the sun has already started the reaction you cannot reverse.
Reef fish, flatfish, cod family and most inshore species are less of a histamine risk, but they carry nematodes in the gut and body cavity. For those, the freezer step matters more than the speed of the ice, because the parasite is the main hazard and the ice is mostly about quality.
The decision order that follows is short. Is the fish a histidine-rich pelagic? Then the ice goes in before anything else, including photos. Are you going to eat it raw? Then the core has to hit the freezing endpoint, and the fish has to be small enough or thin enough for your freezer to get there. Neither? Then normal chilling is enough and you should stop treating the fish like a liability.
That order also explains why sashimi-grade is a bad phrase. It describes a handling record, not a species. A tuna handled badly is not sashimi-grade no matter what the fishmonger's sign says, and a mackerel handled well can be.
What has to happen between the gaff and the freezer?
Bleed it and get it cold. A spike to the brain and a cut behind the pectoral fin, then straight into an ice slurry, is the sequence that buys you the most time. A slurry of ice and seawater holds close to 0°C and makes full contact with the skin, which block ice in a box does not. The kill-and-bleed step itself is covered in ikejime and bleeding.
Do not put a warm whole fish into a freezer and call it done. The Tokyo data shows the core lagging the air by hours even in a lab convection unit. While the outside is freezing, the inside is sitting in exactly the temperature band where histamine bacteria work fastest. Chill first, then freeze.
Label what you froze: species, date, and how you know the core got there. If you used a probe, write the reading. If you used a week in a -18°C drawer, write that instead. The label is what stops a six-month-old fillet from being served raw on the strength of a good memory.
Probe placement is the step that fails quietly. Put the probe in the thickest part, close to the bone, and leave it there for the whole hold if the thermometer allows a remote read. A reading taken at the surface of a frozen fillet only tells you the air is cold, which you already knew.
Then accept the failure mode. A fish that spent three hours in a warm box, or died in the net and sat, is not made safe by a month in the freezer. The freezer handles parasites. It does nothing about a decision that was already lost on deck.
Which catches should not be eaten raw
Skip raw preparation for any pelagic that missed its ice window, any fish that died in the net or the livewell and sat, and any fillet holding visible nematodes. Those worms are usually removed with tweezers and the rest of the fillet cooked, but a fillet you are picking through is not a fillet you serve raw.
Skip it too if you cannot state the core temperature and the hold time. A fridge at 2°C for four days is not a parasite treatment. A freezer that cycles through defrost and never holds below -15°C is not one either, and neither situation is visible from the outside of a zip bag.
Limits and terminology differ by jurisdiction. Singapore's 100 ppm figure is a national limit rather than a global one, and the species most often implicated shift with local fisheries and local handling habits. Use the number as a calibration.
The personal side of the decision belongs to a clinician. If you are pregnant, immunocompromised, or have had stomach surgery, the question of whether raw fish is safe for you is not one this article can answer, and no handling standard substitutes for that conversation.
A restaurant serving raw fish is not working from a different species list. It is working from a documented chain: supplier, temperature log, lot. You can build the same chain at home, and the parts are a slurry, a probe and a label. What you cannot do is skip it and rely on the fish looking good, because both failures in this article are invisible at the counter.
The two checks that cover the rest are simple. Core temperature for the parasite, time from death to ice for the histamine. Get those two right and the rest of the sashimi conversation is about knives and rice, which is not what this piece is about.
Short version
- The freezing endpoint is core temperature: -35°C core, or 24 hours with the core at -20°C. At -18°C, hold for a week.
- Histamine is formed before freezing and survives cooking, so the clock starts when the fish dies, not when it goes in the bag.
- Sashimi-grade describes a handling record. A histidine-rich pelagic that sat warm is not made raw-safe by the freezer.
Sources
- 21 CFR 123.6: Hazard analysis and HACCP plan (fish and fishery products) — US Office of the Federal Register (eCFR) · T1
- Rapid and Conventional Freezing Conditions of Fish for the Prevention of Human Anisakiasis — Food Safety (Tokyo Metropolitan Institute of Public Health) · T1
- Anisakiasis Annual Incidence and Causative Species, Japan, 2018-2019 — Emerging Infectious Diseases (US CDC) · T1
- Two Case Reports of Scombroid in Singapore: A Literature Review — Cureus · T1