Chocolate is dry, which stops bacteria growing in it and does not stop them surviving in it. These are the outbreaks and the measurements behind that distinction.
16 studies1975 to 2026Each checked against its published abstract
What these studies add up to
Salmonella has caused repeated outbreaks through chocolate: a systematic review counted twelve since 1970, with 3,266 patients, most of them children. One was traced to cocoa beans, and in another the bacteria were found in chocolate five months after it was made.
The numbers found in implicated chocolate have been very small: a few cells per 100 g in two outbreaks. Both investigations concluded that a very small dose may have been enough.
It survives for a long time. In one experiment it was still present in chocolate after twelve months at room temperature, and it survived better the drier the chocolate was, and better in milk and white chocolate than in dark.
Dryness also protects it from heat. In dry chocolate and in dry milk powder a tenfold reduction at 80°C took tens of minutes, and in the powder it became far easier to kill once water was added. Heating finished chocolate is therefore not a remedy.
Raw cocoa beans do carry it: about one sample in twelve in the largest survey read here. Roasting can achieve a large reduction and has been measured doing so. Conching does very little.
So the manufacturing problem is what happens after the roaster: bacteria arriving with milk powder, nuts or other inclusions that are never heated again, or from the factory itself. On dry surfaces Salmonella has been shown to form films after only brief wetting, and dry cleaning reduced contamination in a model without eliminating it.
What caused cases of Salmonella Nima infection in Canada and the United States in 1985 and 1986?
How
An outbreak investigation: 29 confirmed cases in Canada and four in the United States, with counts made on four composite samples of the suspect chocolate.
Finding
The cases were traced to foil-wrapped chocolate coins imported from Belgium. The chocolate held 4.3 to 24.0 organisms per 100 g, and the first case in each of 24 affected families was a child of 14 or under who had eaten roughly 25 g.
Does not establish: An infectious dose. The counts are what remained in the product when it was tested, not what the children swallowed.
Limits: Four composite samples, counted after the event, from product that had been stored for an unknown time.
What caused an outbreak of Salmonella Typhimurium infection in Norway and Finland in 1987?
How
An outbreak investigation with laboratory typing of the strain and testing of chocolate taken from shops.
Finding
349 verified cases in Norway and 12 in Finland were traced to chocolate from one Norwegian company. The median age of patients was 6 years. About 90% of positive retail samples held 10 or fewer cells per 100 g.
Does not establish: That fewer than ten cells will make anyone ill. The authors say only that such a dose may have been sufficient here.
Limits: Counts were made on retail samples after the outbreak was recognised.
Where does Salmonella in low-moisture foods come from, how long does it last, and how well does it resist heat?
How
A narrative review of published work on peanut butter, infant formula, chocolate, cereal products and dried milk.
Finding
These foods do not support growth, but Salmonella survives in them for long periods. Contamination has been traced to poor sanitation, poor equipment design and poor ingredient control, and the main risk arises when contamination follows the heating step. Heat resistance generally rises as moisture falls.
Does not establish: Any time and temperature for a process. The review itself warns against applying published heat-resistance values to a product they were not measured in.
Limits: A review, and not a systematic one. It reports what earlier studies found and adds no measurements.
How does Salmonella behave during fermentation, drying and storage of cocoa beans?
How
Pilot-scale inoculation study: fermentation in boxes of 2 kg beans for 7 days with a Salmonella pool, followed by drying and storage at room temperature of beans inoculated during fermentation or during drying.
Finding
After 30 days of storage at room temperature the water activity decreased to 0.68 and reductions of Salmonella ranged from 0.93 to 2.52 log MPN/g. Despite the reductions, the pathogen was detected even after 120 days.
Does not establish: It does not show that stored cocoa beans commonly carry Salmonella, and it does not cover roasting, which is a later control step.
Limits: Deliberately inoculated beans at pilot scale; natural contamination levels are far lower and were not measured.
How effectively do roasting protocols used for fine chocolate inactivate Salmonella on cocoa beans?
How
Whole beans inoculated with one Salmonella strain and roasted at 100 to 150°C for 2 to 100 minutes.
Finding
A reduction of more than 5 log10 after ten minutes at 150°C. The time needed for a tenfold reduction fell from about 33 minutes at 100°C to about 2 minutes at 140°C.
Does not establish: That any particular maker's roast is sufficient. That has to be validated for the equipment and the product.
Limits: A single strain, on whole beans, in a laboratory roaster.
How well does Salmonella survive, and resist heat, in milk powders during storage, dry and after water is added?
How
Nonfat dry milk and whole milk powder inoculated with five serovars and stored for 180 days at about 20°C, with heat resistance measured every 30 days, dry and hydrated, in three replications.
Finding
In dry whole milk powder the time for a tenfold reduction at 80°C rose from 18.9 minutes on day 1 to 29.4 minutes on day 180. Once the powder was hydrated the bacteria were far easier to kill, and were tested at 59 to 65°C instead of 80 to 90°C.
Does not establish: What happens to Salmonella in milk powder once it is inside chocolate, where fat surrounds it.
Limits: Laboratory inoculation at high levels, in powder, not in chocolate.
How does water activity affect the survival and heat resistance of Salmonella in dark, milk and white chocolate?
How
Three serovars inoculated into dark, white and milk chocolate at water activities of 0.25 and 0.50, stored for 12 months at 22°C, with heat resistance measured at 80°C.
Finding
Salmonella was still present after 12 months. Numbers fell more at the higher water activity, and fell furthest in dark chocolate; milk and white chocolate protected it better. In the drier dark chocolate a tenfold reduction at 80°C took 35.7 minutes, against 6.45 minutes at the higher water activity.
Does not establish: That heating finished chocolate is a way of making it safe. The figures show how slowly Salmonella dies in chocolate, which is the opposite point.
Limits: Inoculated in a laboratory at far higher numbers than any outbreak has found in product.
How many outbreaks of salmonellosis have been linked to chocolate, and what do they have in common?
How
A systematic review, registered in advance, of three databases for reports published after 1970 of outbreaks with more than 10 patients.
Finding
23 articles described 12 outbreaks with 3,266 patients in total. All were in high-income countries, five crossed borders, six peaked in winter, and children were the main group affected. No serotype predominated.
Does not establish: That chocolate outbreaks happen only in rich countries. Those are the countries with the surveillance to detect them, which the authors note.
Limits: It counts only outbreaks that were investigated and published, and excludes those of ten patients or fewer.
How often is Salmonella found on raw cocoa beans, and how much does the heating step have to remove?
How
870 samples of raw beans from the main sourcing areas tested over nine months, followed by a risk model from bean to a hypothetical dark chocolate bar eaten in Belgium.
Finding
Salmonella was detected in 71 samples, about 8.2%, mostly at 0.3 to 46 organisms per gram. The model's estimate of illness fell roughly tenfold for each further tenfold reduction achieved in processing.
Does not establish: The risk from a real bar. Every outbreak investigated points to contamination after heating, which this model leaves out.
Limits: The model considers one heating step and, by design, no contamination from the factory afterwards. The origins sampled are not broken down in the abstract.
If a small patch of a dry processing line is contaminated, how much product is affected, and how much do dry cleaning methods help?
How
A simulation of a milk powder line, run 1,000 times, using measured figures for how much a dry towel removes from a surface and how much transfers from surface to powder.
Finding
A small contaminated patch of stainless steel contaminated a median of 72 consumer packs of 300 g. Wiping with a dry towel first brought that to 26. Flushing the line with 150 kg of powder brought the median to 0, with a wide range.
Does not establish: That flushing a line with product makes it safe. The median was zero; the upper end of the range was not.
Limits: A model of milk powder, not chocolate, and only as good as the transfer figures put into it.
Can Salmonella and Cronobacter form biofilms on surfaces that are wetted briefly and then left dry?
How
Laboratory films grown on stainless steel and polypropylene under cycles of a wet phase followed by a dry phase of 48 to 120 hours, examined by culture and microscopy.
Finding
Both formed dry-surface biofilms. How long the surface had been wet mattered most, and some cells passed into a state in which they were alive but would not grow on a culture plate.
Does not establish: How common such films are in factories, or whether cells in that state cause illness.
Limits: Laboratory coupons and chosen strains, with no food residue of the kind a real line carries.
How few cells are enough to cause illness when carried in chocolate. The outbreak counts were made on product after the event, not on what was eaten.
How far roasting results for one strain on whole beans carry over to other strains, to nibs, and to gentler roasts.
How often inclusions and milk powders arrive contaminated. The studies read here measure how long Salmonella lasts once it is there, not how often it is.
Whether dry cleaning can be shown to remove Salmonella from a line. The United States regulator wrote in 2025 that it was not aware of evidence that cleaning alone, including flushing with product, does so.
How common dry-surface films are in working factories, as opposed to on laboratory coupons.
These are questions the studies read for this page leave unanswered. They are not a claim that nobody has answered them.
We would like to set two Google Analytics cookies to count page views, and to count them in our own cookie-free measurement as well. They do not identify you and nothing is shared with advertisers. Decline and the site works exactly the same — see cookies for what each one does.