The blocking works. That is why it does not solve it.
93.57% of blocked domains do not resolve. But 24.3% of the orders target a brand that had already been blocked.
Between January 2025 and July 2026 the Brazilian government ordered 51,150 betting-related addresses to be blocked. Measurement shows the order is carried out in almost every case. The problem lies elsewhere: 24.3% of those blocks fall on a brand that had already been blocked — and the share is rising.
This report uses two bases and never adds them together. Volume, escalation and repeat rate come from the official Ministry of Finance spreadsheet (51,150 dated addresses). Compliance and classification come from the list applied by internet providers (88,997 domains), measured from a single network vantage point in São Paulo. The overlap between the two is 98.8%.
The escalation
Addresses the Ministry of Finance ordered blocked, by quarter. In seven months of 2026 there were 36,831 — 2.55× the whole of 2025.
The peak quarter concentrates 22,532 addresses — 6.5× the first quarter in the series.
The period marked with an asterisk is partial: it covers only July 2026 and should not be compared with a full quarter.
One in four blocks is a repeat
Grouping addresses by brand, 7,738 of the 24,135 orders fell on a brand the government had already ordered blocked. The share is not stable: in the last measured period it reached 39.9%.
198 brands were blocked in three or more separate quarters. Three of them appear in all seven periods of the series — from early 2025 to July 2026.
Compliance is high
Of the 88,997 domains on the list applied by providers, 83,274 do not resolve — 93.57%. The order arrives and is executed.
The measurement compares the provider resolver response with that of a neutral resolver. The signature separating a block from a non-existent domain is in the DNS response itself: the absence of the authority record indicates a synthetic answer from the provider, not a domain that ceased to exist.
This is only observable from inside a Brazilian access network. Blocking is applied at the provider resolver — a probe running in the cloud or in a datacenter network cannot see it.
This measurement was taken from a single vantage point: a residential connection in São Paulo, on 24 August 2026. It describes what happens at that point, not the behaviour of every provider in the country.
Who gets through
4,412 domains on the list still resolve. Of those, 2,403 answered a request — but answering is not operating.
Much of what answers is a shell: 541 addresses return an empty page, a minimal body or a domain-for-sale notice. There is no operation behind them.
After setting those aside, 1,624 betting sites remain effectively running, with sector content identified on the page. Against the betting universe on the list, that is 4.86%.
Classification looks at the content served, not at the domain name. Addresses without a single sector word in the name turned out to be casinos once the page was opened — and the reverse also occurs.
Seven months until the order arrives
Looking up the registration date of domains on the official list, the median between an address being created and being blocked is 209 days. Only 28.8% are blocked within their first 90 days.
Half of the addresses operate for more than 209 days before the block arrives. Combined with the repeat rate, the picture describes the cost of the instrument: the address is blocked, not the operation — and changing address is cheap.
Address rotation, observed
Among blocked domains that keep resolving there is a recurring pattern: the old address stays up and forwards the visitor to a successor not yet blocked, carrying the same session identifier in the URL.
Sequential numbering in the names suggests a queue prepared in advance. 85 destination addresses were identified that were not yet on the blocking list.
Not every redirect is evasion: 31 international addresses of brands licensed in Brazil forward to the authorised domain. There the design is working — the international address is blocked precisely because the legal operation must run on the Brazilian domain.
Methodology
Primary source for volume: response to request 18800.126199/2026-92, addressed to the Brazilian Ministry of Finance under the Access to Information Act (Law 12.527/2011), granted on 10 August 2026. The attachment lists 51,295 rows with a submission date, across seven quarterly tabs, corresponding to 51,150 unique addresses. The source labels the column as URLs; in practice the entries are host names, and only five of them carry a path.
Base check: the January 2025 to June 2026 window totals 45,030 addresses against 45,037 stated in the earlier response — a difference of 7 rows. Comparing the two responses month by month, every month matches except three: February 2025 is 7 rows short, and February 2026 and May 2026 differ by one row each, in opposite directions. Both responses come from the same body and the divergence is recorded as found, without interpretation.
Applied list: public file maintained by the community of internet providers, with 88,997 domains as of 24 August 2026. Its overlap with the official list is 98.8%, which allows treating it as a mirror of the official source for measurement purposes.
Compliance measurement: DNS query for each domain on the applied list, comparing the provider resolver response with a neutral resolver, and using the presence of the authority record to distinguish a block from a non-existent domain.
Content classification: for domains that answered, reading of the served page and of the JavaScript bundle it loads, looking for sector vocabulary, game supplier names and the responsible-gaming legal notice.
Time until blocking: domain registration lookup for a random sample of 900 addresses from the official list, of which 524 had a recoverable registration date.
Brand grouping: domain name root with digits removed, discarding generic sector roots. Consolidation, grouping and calculation: VerificaBet.
What this data does not say
The compliance measurement comes from a single network vantage point — a residential connection in São Paulo, on 24 August 2026. A domain that does not resolve there may be reachable on another provider. The report describes what it measured and does not claim national behaviour.
The two bases are never added together. Volume and repeat rate come from the official spreadsheet; compliance and classification from the applied list. Each figure states its origin.
Adding up the bars is not the address count. 145 of the 51,295 records appear in more than one quarter, which yields 51,150 unique addresses. One of the spreadsheet tabs also carries an internal label that differs by one row from that tab's contents; this report uses the row count and records the difference without interpreting it.
Grouping identifies a brand, not a company. The naming pattern strongly suggests the same operation but does not establish corporate ownership: attempts to recover company identification from public surfaces produced no verifiable result.
No brand is named. Some roots of blocked domains coincide with names of brands licensed in Brazil, and a named ranking without proper separation would attribute irregularity to an authorised operator.
The applied list has mixed scope. Besides betting, it includes court orders on broadcast piracy and other objects. Only the subset classified as betting is used in the escape figures.
Content classification is heuristic. It supports the published aggregate and does not sustain a claim about an individual address.
The last period in the series is partial, covering July 2026 only.
The repeat rate is not attributed as a failure of any authority. It describes the cost of the instrument when facing an actor that changes address.
Blocking orders for betting domains
Frequently asked questions
How many betting sites has Brazil blocked?
Between January 2025 and July 2026 the Ministry of Finance recorded 51,150 blocked betting-related addresses. In 2026 alone, over seven months, there were 36,831 — 2.55× the total for all of 2025.
Does betting site blocking work in Brazil?
93.57% of the 88,997 domains on the applied list do not resolve from a residential connection in São Paulo, as measured on August 24, 2026. Compliance with the order is high. The limit of the instrument shows elsewhere: 24.3% of orders fall on a previously blocked brand.
How many blocked sites are still online?
4,412 domains on the list still resolve. After discarding empty pages and domains for sale, 1,624 are betting sites effectively in operation — 4.86% of the betting universe on the list.
Why does a blocked site come back under another address?
Because blocking targets the address, not the operation. The report identified 85 destination addresses receiving traffic from blocked domains and not yet on the list. Sequential numbering in the names suggests a queue registered in advance.
How long does it take for a betting site to be blocked?
The median between domain registration and the blocking order is 209 days, in a sample of 524 addresses. Only 28.8% are blocked within their first 90 days of existence.
Where does the data in this report come from?
Volume and repeat rate come from an official Ministry of Finance spreadsheet obtained through a freedom of information request. Compliance comes from network measurement carried out by VerificaBet over the list applied by providers, from a single vantage point in São Paulo. The two bases are never added together.