Eating Street Food Without Getting Sick

The most quoted piece of travel advice in the world is boil it, cook it, peel it, or forget it. It appears exactly once in the current edition of the CDC's Yellow Book, in the past tense, in a sentence written to explain why it is no longer the guidance. What the CDC says instead is that people who follow those rules still get ill, and that the biggest contributors are hygiene practices in local kitchens and the sanitation infrastructure underneath them.

That is not permission to stop caring. It is a redirection: the thing to watch is not what category of food you are eating, but how long it has been sitting at the wrong temperature. This page sets out what the health agencies actually publish, what the numbers actually are, and which of the famous rules have evidence behind them.

We are a food tour site, not a medical one. Everything below is sourced to the World Health Organization, the CDC, the UK's national travel health service, or peer-reviewed studies, and none of it replaces advice from a travel clinic before you go.

The World Health Organization's Five Keys to Safer Food are the backbone of every national guideline, and they are worth reading because they are about temperature and time rather than about categories of food. Keep clean. Separate raw and cooked. Cook thoroughly. Keep food at safe temperatures. Use safe water and raw materials.

The numbers attached to them are the useful part. Cook to 70 degrees. Do not leave cooked food at room temperature for more than two hours. Keep hot food above 60 degrees until it is served. And the range in between, from about 5 to 60 degrees, is what the WHO poster labels the danger zone. Almost every decision on this page follows from that band.

The CDC's position on street food is blunter than people expect. It states that eating food from street vendors increases the risk of illness, and its public-facing page says to avoid it. We are not going to pretend that says something else. But the same CDC also states that fully cooked foods served hot are the safest, and it is that sentence, rather than the category, that does the work.

The most useful formulation we found belongs to the UK's national travel health service, because it is conditional rather than prohibitive. It advises avoiding food from street traders unless it is thoroughly cooked in front of the traveller and served hot on clean crockery. That single sentence is a usable test, and it is close to a description of a well-run food tour.

We went looking for evidence that street food is unfairly maligned, and the picture is more interesting than either side of that argument.

On one hand, the CDC does say the risk is higher, and there are studies showing poor hygiene at street level. On the other, three things point the other way. In the European Union's food-borne outbreak reporting for 2024, of 533 outbreaks with strong evidence and a known setting, 170 were traced to a restaurant, bar, hotel or caterer and 5 to a street vendor or market. The prospective cohort work on traveller's diarrhoea identifies age, trip length, destination and travel style as risk factors, and does not identify the type of establishment as one. And Bangkok's own municipal sampling, which is the only programme we found that tests seven types of premises with one method, puts street vendors at 19 to 23 per cent non-compliance and restaurants over 200 square metres at 21 per cent: the same band, with the worst performer changing from year to year.

Two caveats that stop this being a knockdown. Outbreak counts are not risk per meal, because nobody knows how many meals were eaten in each setting. And the EU figures come from a region where street food is a small share of eating.

So the defensible claim is the narrow one, and the WHO reached it back in 1996 when it wrote that a stationary vendor serving full meals should be regulated much like a small restaurant. The counter is not the risk. The clock is.

You will see thirty to seventy per cent quoted everywhere for the chance of getting traveller's diarrhoea. That range comes from an airport survey conducted in 1996 to 1998, and it is the CDC's own historical figure.

The current best estimate is lower. A 2025 pooled analysis of ten cohorts covering 8,478 travellers from high-income countries put the incidence at about 36 per cent, and found it falling: 31.5 per cent in studies from the last decade against 40.1 per cent in older ones.

Where you go matters far more than what you order. The UK service groups destinations into three bands: up to 7 per cent for western Europe, the United States, Canada, Australia, New Zealand and Japan; 8 to 20 per cent for southern Europe and parts of the Caribbean and Pacific; and over 20 per cent for Asia, the Middle East, Africa and Latin America. Of the sixteen cities on this site, that puts Tokyo, New York and Melbourne in the lowest band, Rome, Florence, Naples, Porto, San Sebastian and Athens in the middle, and Hanoi, Hoi An, Bangkok, Mexico City, Lima, Medellin and Marrakech in the highest.

These are the signals with published evidence behind them, rather than the ones that get repeated.

Follow the queue of locals. There is no study, in either direction. There is research showing that customers choose on the appearance of the food and the vendor, while vendors concentrate on appearance and neglect the practices that matter during preparation. A queue tells you the food tastes good and the price is right, which are excellent reasons to join it and not evidence of anything else.

Gloves mean hygiene. Contradicted three times over. Workers wearing gloves attempt handwashing about half as often as workers without them. A trial across hundreds of tortillas found no measurable microbiological benefit. And in recent United States outbreak data, contact with ready-to-eat food by a gloved hand was the leading contributing factor in viral outbreaks. A glove that is never changed is a surface that is never washed.

Never buy from someone who handles money. The research on banknotes is all detection and no transmission: nobody has demonstrated the route from note to food, and no outbreak has been attributed to it. The signal that matters is not whether they touch cash but what they do next.

Reheating makes it safe. In Dakar, reheating previously cooked food came out as a risk factor rather than a protective one, with roughly five times the odds. Reheating rescues food that was cooked properly and cooled properly. It does not rescue food that spent the afternoon at thirty degrees.

One more, because it turns up in a great many articles: the claim that a specific number of billions of people eat street food every day. We tried to source it. The citation leads to a page that no longer exists and the figure appears in none of the underlying documents. It may well be true. Nobody can show you where it comes from.

Tap water is where a general rule fails hardest, so here is what we could confirm from the water authorities themselves rather than from a travel guide.

Ice is not a smaller version of the same question. Freezing does not kill bacteria: it suspends them, and they revive as the ice melts. The CDC puts the condition on the water rather than on the ice, and adds, for anyone hoping otherwise, that the alcohol in a drink will not kill anything in ice made from contaminated water.

The commercial-ice distinction is real and it is measured. In one Thai study, 29 per cent of ice sampled from street carts carried resistant Enterobacterales while all ten samples of packaged commercial ice were clean. But packaged is not a guarantee either: a survey of packaged ice across twelve Mexican states found nearly half outside the official guidelines, and a Vietnamese provincial survey found over half of production contaminated. Tubular ice with a hole through the middle is usually the commercial kind, and it is a better bet than crushed ice of unknown origin. It is not a certainty.

Two vaccines are relevant to food and water, and the recommendations differ by destination. Hepatitis A is recommended for unvaccinated travellers to Vietnam, Thailand, Mexico, Morocco, Colombia and Peru. For Greece, Spain and Japan the CDC says to consider it, and its notes specifically mention travellers who plan on eating street food. Typhoid is recommended for most travellers to Vietnam, Thailand, Mexico and Morocco, and worth considering for Peru and Colombia if you are going anywhere with poor sanitation.

Cholera vaccination is recommended for none of the thirteen countries on this site. And the CDC repeats a warning worth carrying: the typhoid vaccine is not fully effective, so it is a reduction in risk rather than a licence to stop paying attention.

One non-food item, because it catches people out on this exact circuit: Colombia requires yellow fever vaccination for arriving travellers, and Peru recommends it for the low-altitude east but not for Cusco or Machu Picchu. Check the current requirement before you fly, because entry rules move.

New York grades restaurants A, B or C on an inspection score, and has done since 2010. The part almost nobody knows changes how you read them: a B or a C is never the result of a first inspection. A premises that fails its initial visit is scored but ungraded and gets a second chance at least a week later. Looking at the city's own published inspection data for this year, initial inspections produced twelve and a half thousand A grades and not a single B or C. Every B in a window is a second attempt, after roughly two thirds of the first-round failures had already recovered an A.

And the rule does not apply to carts. Mobile food vending units are explicitly excluded from the grading section, so the hot dog cart and the halal cart are licensed and inspected but carry no letter and generate no public score.

Does grading work? The famous Los Angeles study that started the practice has been undermined: a re-analysis found the same apparent effect in counties that never introduced grading at all. Better evidence survives elsewhere, including a measured fall in salmonellosis in New York relative to the rest of the state, and a national comparison in which posting results at the door was associated with far fewer outbreaks than posting them online or not at all.

Elsewhere on this site, the picture is patchier. Japan has required food safety management for all operators since June 2021 but publishes only violators, in Japanese. Fukuoka is a striking exception, with a city ordinance that publishes compliance for each individual street stall. Thailand certifies premises but its lists live in a shared drive in Thai, so a visitor sees a sticker and nothing else. Vietnam has ten published criteria for street vendors, and when researchers measured compliance in one city, only four of the ten were met by more than half of vendors.

Most episodes are self-limiting and short. The priority is fluid and salt replacement rather than stopping the symptom, and the WHO publishes a rehydration formula you can make from what is in any shop: six level teaspoons of sugar and one level teaspoon of salt in a litre of safe water. Sachets are easier and worth carrying.

Seek medical help rather than self-treating if there is blood, a high fever, signs of dehydration, or if it goes on beyond a few days, and always for a young child or an older traveller. And take seriously the point above about antibiotics: current guidance is against them for mild cases, and the resistance you can pick up travels home with you.