What Are the Chances of Being Robbed in Brazil?

Why the question has no answer for you

This is worth stating before any number: Brazil's crime statistics do not carry nationality as a variable. Not the federal dataset, not the Public Security Forum's annual, not the state-level open bases.

So a "chance of being robbed as a tourist in Brazil" cannot be computed from Brazilian sources. Where you see one, it has been derived — usually by taking a national rate and presenting it as a personal probability, which is a different claim.

The information does exist administratively in at least one place: Rio operates a police station dedicated to crimes against tourists. What it does not do is publish a series. So the correct statement is not "Brazil does not know", it is "Brazil does not publish it openly".

The most common property crime is fraud, not robbery

Here is the number that should reorganise your precautions. In 2025 Brazil recorded over 2.26 million fraud cases — up from about 427,000 in 2018, growth of more than 400% in seven years.

Compare that with roughly 309,000 phone robberies in the same year. Fraud is not merely larger; it is an order of magnitude larger, and it is the crime Brazilians themselves fear most: 83.2% of the population say they fear being defrauded online or by phone.

For a visitor this reframes the risk. The scenario your instincts prepare for is a confrontation on a street. The scenario the data points at is a screen — a fake payment request, a cloned checkout, a message that looks like your bank. Which is why the most useful safety reading here is about how payment actually works: PIX and what it does, and why a cashier asks for your tax number, because knowing what a normal transaction looks like is what makes an abnormal one visible.

Theft against robbery: the distinction that matters

In Brazilian statistics roubo means taken with force or threat; furto means taken without. English collapses both into "theft" and loses the only part that affects your safety.

For phones in 2025: about 484,000 thefts against 309,000 robberies. Theft leads by more than half again.

There is a second layer in the handset counts. 831,000 phones were taken across 792,000 incidents — more devices than events, because a single robbery sometimes takes several phones at once.

Age patterns differ between the two, and not in the direction you would guess: young people aged 15–29 account for 37.7% of robbery victims, while 45.2% of theft victims are aged 20–39. Robbery skews younger than theft.

Geographically, the highest phone-crime rates are not in thesouth or southeast: São Luís and Belém lead the country, with São Paulo the only southeastern capital near the top. If your itinerary is Rio and São Paulo, the national ranking is not describing your trip.

Roughly half of it never gets counted

Every figure above comes from police records, which means it counts only what somebody reported.

Brazil measured that gap. A national victimisation survey for 2021 found that, for incidents outside the home, victims went to the police in 44.8% of thefts and 57.9% of robberies. So somewhere around half of ordinary property crime never enters any statistic — and for financial fraud the under-reporting is worse still.

Two consequences. Real incidence is meaningfully higher than the published rates, in a way the published rates cannot show. And Brazil has no periodic victimisation survey — these run years apart — so the correction factor is itself dated between rounds.

Where the granular data exists, and how it misleads

São Paulo publishes something remarkable: a street-level file of every phone reported stolen, with coordinates, time, and the device brand — hundreds of thousands of rows for 2025 alone.

It comes with warnings from the state itself, and they matter more than the file. The data is explicitly not the official statistic — the regulation defining São Paulo's criminal statistics does not include it. Around 60% of reports are filed at a police station away from where the crime happened, so mapping by station location maps the paperwork, not the crime. And the row count is not an incident count: supplementary filings repeat, and the state says so.

Anyone who has seen a heat map of phone theft in São Paulo has probably seen one built on those rows without those caveats.

For the geographic picture rather than the probabilistic one, what the national numbers say and Rio measured neighbourhood by neighbourhood are the closer answers, and what your own government publishes is worth reading with its date in view. Practically: carry less — cash without a local card, which ATM to use, what you can bring in — and keep a working local SIM, since a phone is both the thing most likely to be taken and the thing every mitigation depends on.

Sources

• Anuário Brasileiro de Segurança Pública 2026 (FBSP) — phone robbery and theft by state and city, handset counts, age profile, the fraud series, and the fear figure. Accessed September 3, 2026.
• Mapa de Segurança Pública 2026 (Ministry of Justice) — the list of indicators states are obliged to report, which is what establishes that phone crime and express kidnapping are not among them. Accessed September 3, 2026.
• PNAD Contínua victimisation survey — reporting rates of 44.8% and 57.9%. Accessed September 3, 2026.
• SSP-SP phone-crime microdata and its accompanying limitations note — the street-level file, the warning that it is not the official statistic, and the 60% figure for reports filed away from the scene. Accessed September 3, 2026.

What we could not verify

• Any probability for a visitor. No Brazilian dataset records nationality. Rio's tourist-crime station holds the administrative data; it is not published as a series.
• How much of the fraud total involves foreign victims or foreign cards. The category is reported as a count without that breakdown.
• A current national under-reporting rate. The most recent measurement is for 2021, and there is no periodic survey to update it.
• Whether the share of victims who suffer physical aggression during a robbery still holds. That figure circulates widely but traces to a survey from the early 2010s, and we found no recent national measurement of it.