Phone scam and robocall statistics 2026

Every number on this page comes from a named public source: the Federal Trade Commission's fraud reports and Do Not Call data, the FBI's Internet Crime Complaint Center, the Federal Communications Commission, and YouMail's robocall estimates. We also read this year's FCC complaint records ourselves, for three things no agency reports: when scam calls come, how rarely a caller's number is reported twice, and where caller IDs fake a local number most. Each figure carries a numbered link to its source, and the whole dataset is available as a CSV at the end of the page.

How much money do people lose to phone scams?

People told the FTC they lost $1.1 billion² in 2025 to fraud that began with a phone call, across 290,429² reports. That is a small slice of all reported fraud, which came to $15.9 billion¹ from 3 million¹ reports. It is also a slice that has grown every year: phone-call losses rose 39.5%² from 2022 to 2025, while the number of phone-call reports hardly moved.

The typical loss is rising too. The median phone-call loss was $1,400² in 2022 and $1,835² in 2025, and it reached $1,997² in the first half of 2026. The share of phone-call reports that involved a loss climbed alongside it.

Phone calls vs texts, email and social media

Which channel is worst depends on what you count. Text messages start the most scams: 411,438² reports in 2025, against 290,429² for phone calls. Social media takes the most money in total, $2.1 billion³. But a phone call is the most expensive way to be scammed. The median loss on fraud that began with a call was $1,835², higher than for any other contact method.

Calls also turn into losses far more often. 19%² of phone-call reports included a money loss, against 9%² of text reports. So phone calls produced 1.7x² the losses of text messages, from fewer reports.

Which scams arrive by phone?

Impersonation dominates. Among scams that began with a phone call in 2025, callers posing as a government agency took the most money, followed by callers posing as a business.

Government impersonation is growing well beyond the phone. Across all contact methods, government-imposter reports nearly doubled from 2022 to 2025, while the median loss fell from $1,500 to $645: more attempts, each taking less. The FTC puts all imposter losses at $3.5 billion⁴ in 2025, including about $920 million⁴ lost to government impersonators.

The FBI's view of call-center fraud

The FBI sees the same pattern. Its Internet Crime Complaint Center groups tech-support and government-impersonation scams as call-center fraud, and counted more than 80,000¹⁰ such complaints in 2025 with losses above $2.9 billion¹⁰. Government-impersonation losses reported to the FBI rose 96.7%¹⁰ in a single year.

Who loses the most to phone scams?

Age changes the picture more than anything else. Across all fraud in 2025, people 80 and over were the least likely to report losing money, at 22.2%² of their reports, but had the highest median loss, $2,000². People in their twenties lost money more often and lost far less, a median of $305².

The phone matters most at the oldest ages. The FTC found that for people 80 and over, phone calls cost more than any other contact method, while social media cost the most for every younger age group. For people 60 and over, the median loss on fraud that began with a call was $3,000⁵ in 2025. Among older adults who lost $10,000 or more to a business or government imposter, 41%⁶ said a phone call came first.

The FBI's numbers for people 60 and over are larger again: 201,266¹⁰ complaints and $7.7 billion¹⁰ in losses in 2025, up 59%¹⁰ from 2024. 12,444¹⁰ people in that age group reported losing more than $100,000.

How many robocalls do Americans get?

YouMail, which makes a call-blocking app, estimates that 3.8 billion¹¹ robocalls reached US phones in September 2026. That is about 128 million¹¹ a day and 11.7¹¹ for every person. Its estimate for all of 2025 is 52.5 billion¹¹.

Volume is falling, slowly. January to September 2026 ran 10.4%¹¹ below the same months of 2025, and the latest twelve months add up to 48.3 billion¹¹. The peak year in the index was 2019, at 58.5 billion¹¹.

When do scam calls come?

In office hours. We read the time of every unwanted call reported to the FCC this year that gave one: 93%¹⁴ came between 8 a.m. and 9 p.m., the only hours telemarketing law allows, and 92%¹⁴ came Monday to Friday. Sunday got 2.8%¹⁴.

That is the shape of a working week, not of a machine left running. Many robocall campaigns hand anyone who answers to a live agent, so the calls run while call centers are staffed. Calls that break every other rule still keep to the law's calling hours.

Do Not Call complaints are rising again

Robocall volume and robocall complaints are moving in opposite directions. Do Not Call complaints to the FTC rose to 2.6 million⁸ in fiscal 2025, and complaints about robocalls jumped 45.7%⁸ to 1.6 million⁸, or 61.2%⁸ of the total. Complaints are still about 48%⁹ below fiscal 2021.

The robocalls people complain about are mostly selling debt relief.

The registry keeps growing, to 258.5 million⁸ active numbers. Registration only binds callers who follow the law; scammers ignore it.

The FCC hears the same rise. It received 125,199¹⁴ unwanted-call complaints in 2025, up from a low in 2023.

Why a local number proves nothing

Caller ID is easy to fake. The largest illegal robocall operation the FCC has investigated placed more than 5 billion¹² auto-warranty calls in three months of 2021, displaying 1,051,461¹² different caller IDs. About 75%¹² of the caller IDs in related complaints used the consumer's own area code, a neighboring one, or one in the same state, a practice the FCC calls neighbor spoofing. The FCC fined the operation $299,997,000¹², its largest robocall penalty.

This year's complaints show how little a caller ID number is worth. Of the 45,291¹⁴ different caller ID numbers in unwanted-call complaints to the FCC from January to September 2026, 97%¹⁴ were reported only once. Callers rotate or fake their numbers, so blocking the one that just called you stops almost nothing.

How local the fake looks depends on where you live. In Colorado, 70.9%¹⁴ of complaints that gave a caller ID showed a Colorado area code, against 41.5%¹⁴ nationally, and the share jumped in March. An in-state number is no reason to pick up.

Tracing calls back to their source is slow work. In 2024 the industry's traceback group started 3,606¹³ tracebacks of suspected unlawful robocalls. 69%¹³ of completed tracebacks led the originating provider to warn or cut off the caller, down from 84%¹³ in 2023.

How we compiled these statistics

Every figure was read in its source on the retrieval date printed under each chart and table, and each one links to the source list at the end of the page. We used no figure that appears only on a statistics roundup site.

As of mid-October 2026 the FTC had not published its 2025 Data Book, so the 2025 FTC figures come from its congressional testimony, its 2026 releases, and its public Consumer Sentinel dashboard, with data as of 30 June 2026. We read the dashboard through its own year filters. To check that reading, we compared the dashboard's 2024 phone-call figures with the printed 2024 Data Book: 284,659⁷ reports, $948 million⁷ in losses and a $1,500⁷ median in print, against 284,613 reports and the same losses and median on the dashboard today. The small difference in reports is the FTC adding late reports.

Figures marked as derived are our arithmetic on published numbers. The CSV names the inputs for each one.

What these numbers can't tell you

• Reports are self-selected. The FTC and FBI count what people choose to report, and most fraud is never reported.
• Contact method means first contact. It records how a scam first reached someone, and many reports do not name one.
• The FTC and FBI datasets overlap. Never add them together.
• Do Not Call data runs on fiscal years, October to September.
• Robocall volumes are YouMail estimates, not counts.
• Dashboard figures are revised as late reports arrive, so a figure read later may differ slightly.