Cereulide and Bacillus cereus: Why Reheating Will Not Help

Cereulide is the vomiting toxin made by some strains of Bacillus cereus. Once it is in the food, cooking will not take it out. Hong Kong’s Centre for Food Safety reports that it survives 126 °C for 90 minutes, far more heat than any reheating step gives it.

That one property explains two very different headlines. In 2026, recalled infant formula batches reached 99 countries and territories after cereulide was traced to an ingredient oil, and South Africa’s National Consumer Commission listed two formula recalls linked to the same toxin. In kitchens and factories here, the same toxin is behind the classic case of cooked rice left out too long, reheated until steaming, and still making people sick within hours.

The control is time and temperature before the toxin forms, and the right test when you need proof. If you want your own cooling step or ingredient specifications checked, ask our food science team for a free, no obligation quote.

What happened with infant formula in 2026?

According to Nestlé’s own timeline, routine checks at a Dutch factory found very low levels of cereulide at the end of November 2025. The source was traced in December to an oil blend containing arachidonic acid (ARA) oil from a global ingredient supplier. Germany’s risk assessment institute, the BfR, describes the ARA oil as derived from the fungus Mortierella alpina. Other formula makers that used the same oil recalled batches over the following weeks.

The World Health Organization counted 144 confirmed and suspected cases in ten countries by 25 February 2026, and reported that recalled batches had gone to 99 countries and territories.

On 2 February the European Food Safety Authority set an acute reference dose of 0.014 micrograms per kilogram of body weight for infants. For a young baby on reconstituted infant formula, EFSA put the level of concern at about 0.054 micrograms per litre, a fraction of a millionth of a gram in a whole litre of feed.

In September 2026 the European Commission consulted on a legal maximum level for cereulide in dried infant formula and related infant foods, to sit in Regulation (EC) 2073/2005. South African suppliers to EU formula makers should expect cereulide results in their contracts.

South Africa’s National Consumer Commission recall list shows NAN Special Pro HA, received on 5 January 2026, for the potential presence of cereulide, and Aptamil Nutribiotik 2 and Aptajunior Nutribiotik 3, received on 4 March 2026, for a raw material carrying traces of cereulide. Once the toxin is present, nothing at the end of the line removes it. Control has to sit with the supplier, the process and the cooling step.

Why does reheating not make the food safe?

B. cereus lives in soil and dust and arrives on rice, maize meal, flour, spices, dried vegetables and dairy powders as spores. Spores survive normal cooking. When the food cools slowly, the spores germinate and the cells multiply.

According to the Food Safety Authority of Ireland factsheet, the organism grows from about 4 °C to 55 °C, fastest between 30 °C and 40 °C, at pH 4.3 to 9.3 and down to a water activity of 0.93. Cooked rice, pap, pasta and sauces sit comfortably inside every one of those limits.

Emetic strains make cereulide while they grow in the food. Reheating to 75 °C kills the vegetative cells, so the count can drop sharply while the toxin is untouched. The WHO notes that cereulide stays active through the acid of the stomach and that there is no antidote. Symptoms are nausea and vomiting, usually within half an hour to six hours.

Because the illness is short, most cases never reach a clinic and are blamed on “something I ate”. The other type, the diarrhoeal illness, is different. There the toxins are made in the gut after large numbers of cells are eaten, and onset comes later. Both are prevented the same way: do not give the spores time at warm temperatures.

What makes cereulide so hard to remove?

For technologists, the chemistry explains the behaviour. Cereulide is a small ring shaped peptide, a cyclic depsipeptide of about 1.2 kDa, assembled by an enzyme complex coded by the ces genes of emetic strains. It is not a protein in the usual sense, so heat, pH from about 2 to 11 and digestive enzymes leave it intact. The EFSA BIOHAZ opinion on Bacillus cereus reviews the evidence.

It is also strongly fat soluble. In the body it carries potassium ions across membranes, damages mitochondria and triggers vomiting through serotonin receptors on the vagus nerve. In a factory, fat solubility means the toxin follows the oil phase. That is how an oil ingredient could carry cereulide with no live cells left to find.

How fast must cooked rice and starchy food be cooled?

Regulation R638 sets holding temperatures in Annexure E: chilled at 5 °C or below (4 °C for raw fish, poultry, offal and milk) and hot at 60 °C or above. It does not set a cooling time for cooked food. So many South African HACCP plans adopt an international guideline. Either of these is defensible if the plan names it, validates it and monitors it.

• Codex catering code (CXC 39-1993, formerly CAC/RCP 39-1993), section 7.7.2: cool from 60 °C to 10 °C at the centre in less than two hours, then store at 4 °C. Under section 7.10, reheat to at least 75 °C at the centre within one hour of taking the food out of refrigeration, and keep it at 60 °C or above until served.
• US FDA Food Code, section 3-501.14: cool from 57 °C to 21 °C within two hours, and from 21 °C to 5 °C or below within a further four hours. It is a US model code, not South African law, but its first stage covers the fastest growth range.

The physics is simple. A deep pot of rice holds its heat in the centre for hours. Shallow trays, food no deeper than about 5 cm, a blast chiller or an ice bath, and lids left off until the food is cold do most of the work.

A fridge is built to keep cold food cold. It is not built to pull 35 litres of hot rice down in two hours, and loading it with hot food also warms everything else on the shelves.

Worked example: validating rice cooling in a Gqeberha ready meal kitchen

Take a kitchen that cooks rice to give 30 kg cooked weight at 11:00, for chilled meals packed the next morning. Until now the rice has been cooled in two 20 litre pots in the walk-in cold room, with a single temperature check at 16:00. The HACCP team adopts the Codex limit and runs a validation, logging the centre temperature with a probe logger every 15 minutes.

• Old method, two deep pots in the cold room: the centre is still above 85 °C after two hours, has not reached 60 °C by late afternoon, and is still above 10 °C when the packing shift starts the next morning. It fails badly, and spends most of the night in the growth range.
• New method, 65 mm gastronorm trays filled no deeper than 50 mm, uncovered, in a blast chiller: the centre reaches 10 °C in 85 to 100 minutes from 60 °C on each trial. It passes with margin.

The team runs the trials under worst case conditions: the largest batch, a fully loaded chiller and a hot afternoon. Tray depth and trays per cycle go into the HACCP plan as validated conditions, with the logger files attached.

Day to day, the cook records the time cooking ends, the time and centre temperature at loading (60 °C or above), and the centre temperature two hours after loading, with a probe checked against a reference thermometer every week. Rice above 10 °C at two hours is discarded, not reheated, and the deviation is recorded with its cause.

These figures illustrate the method; your own trays, chiller and batch size decide your result. Our article on cooking, cooling, reheating and hot holding CCPs shows a monitoring sheet built on the FDA two stage limit. Either limit works, provided the plan names one and validates it. HACCP verification vs validation explains the difference between this trial and routine checks.

What should a factory ask its ingredient suppliers?

The formula recalls started with an ingredient, not a process step at the formula plant. A certificate of analysis that reports “B. cereus less than 100 per gram” tells you about live cells on the day of testing. It says nothing about toxin made earlier in the supplier’s process, and for a refined oil a count tells you nothing at all.

Even the EU microbiological criterion for dried infant formula for babies under six months, under Regulation (EC) 2073/2005, 50 to 500 presumptive B. cereus per gram, is a count and not a toxin test.

1. Rank ingredients by risk. Starch based, fermented or slow dried ingredients, and oils made by fermentation such as ARA and DHA oils, going into foods for infants, the elderly or the sick come first.
2. Ask how the supplier controls growth of B. cereus in its own process: times, temperatures and holding steps, not only the final count.
3. Specify the right test. For high risk ingredients, ask for cereulide results by ISO 18465, the LC-MS/MS method for cereulide, from a laboratory accredited for that method. Check the SANAS schedules, and if no local laboratory is accredited for it, name an accredited laboratory abroad in the specification. Presumptive counts to ISO 7932 stay useful for trending.
4. Write the limit into the specification, with the method, the sampling plan and what happens to a lot that fails.
5. Audit the high risk suppliers. A desk review of certificates is not enough for an ingredient that could put your brand on a recall list.

Want our food science team to rank your ingredients and write cereulide into your specifications? Get a food science quote. Our supplier approval risk assessment guide shows how to rank suppliers, and what to test and how often covers choosing and briefing a laboratory.

What does this mean for product development and shelf life?

• Shelf life trials for chilled rice, pasta and pap meals should include an abuse temperature, such as 7 to 8 °C in a retail chiller, with B. cereus counted at the end of life.
• Line intermediates such as soaked grains, starch slurries, batters and custard bases held warm on the line are a cooling step too, and need the same validation.
• Acid as the control, as in sushi rice, only works if the recipe specification keeps the pH below the 4.3 growth minimum, confirmed by measurement and sensory work.

How big a problem is Bacillus cereus in South Africa?

An NICD analysis of foodborne disease outbreaks in South Africa from 2011 to 2020 counted 892 outbreaks. Where a pathogen was identified, B. cereus was found in 31% of them (99 of 319), second only to non-typhoidal Salmonella at 34%. Households were the most common setting, at 35% of outbreaks.

Summer makes it worse: hotter kitchens, cold rooms working at their limit and big batches cooked hours before a function. A cooling step that passed in July can fail in December, so the time to re-check it is now.

Questions we are often asked

Does reheating rice kill Bacillus cereus?

Reheating to 75 °C kills the growing cells, but it does not destroy the cereulide toxin they have already made, and it does not kill the spores. Rice that has been held warm for hours is not made safe by reheating. Cool it fast after cooking and keep it at 5 °C or below.

How long can cooked rice stay at room temperature?

No longer than it takes to cool it. Codex guidance for catering is to cool cooked food from 60 °C to 10 °C at the centre in under two hours. Rice left out longer than that should be thrown away, not reheated.

Can a laboratory test food for cereulide in South Africa?

Cereulide is measured by LC-MS/MS, and ISO 18465 is the standard method. Ask the laboratory whether it is accredited for that method before you send samples. A presumptive B. cereus count to ISO 7932 is cheaper and useful for trends, but it does not measure the toxin.

Was South Africa affected by the 2026 infant formula recalls?

Yes. The National Consumer Commission lists NAN Special Pro HA (received 5 January 2026) for the potential presence of cereulide, and Aptamil Nutribiotik 2 and Aptajunior Nutribiotik 3 (received 4 March 2026) for a raw material carrying traces of cereulide. The WHO’s case count for the event did not include South African cases.