Research · International trademarks
The USPTO provisionally refuses 96% of international trademark designations. Only 2.6% stand.
The U.S. is the hardest major office to enter through the Madrid System, on paper. Of 44,147 designations of the United States registered 2019–2022, 95.9% drew a provisional refusal. Yet only 2.6% of those designations are refused today, and 80.2% hold a grant of protection. The provisional refusal is the USPTO's office action, and almost everyone gets one.
The number that scares Madrid filers
Ask a foreign trademark owner about extending their international registration to the United States and you will hear the same warning: expect a refusal. The data says the warning is right. Among the 53 offices that examined at least 5,000 designations in our 2019–2022 cohort, the USPTO ranks #1 for provisional refusals at 95.9%. The European Union office (EUIPO) sits at 20.3%, the United Kingdom at 11.4%.
| Designated office | Provisional refusal rate | Provisional | Still refused | Protected | Designations |
|---|---|---|---|---|---|
| United States (US) | 95.9% | 2.6% | 80.2% | 44,147 | |
| Thailand (TH) | 92.8% | 49.7% | 47.9% | 22,091 | |
| Canada (CA) | 85.6% | 19.6% | 77.1% | 35,712 | |
| Philippines (PH) | 73.7% | 4.7% | 53.5% | 17,557 | |
| South Korea (KR) | 68.9% | 4.9% | 81.7% | 28,579 | |
| Japan (JP) | 63.6% | 14.3% | 85.0% | 32,759 | |
| China (CN) | 61.7% | 48.8% | 50.8% | 40,684 | |
| India (IN) | 54.9% | 23.5% | 73.6% | 28,230 | |
| Israel (IL) | 53.3% | 7.9% | 89.5% | 13,370 | |
| Cambodia (KH) | 43.2% | 19.5% | 77.0% | 6,714 | |
| Russia (RU) | 40.4% | 11.7% | 88.2% | 28,153 | |
| New Zealand (NZ) | 37.4% | 2.3% | 94.9% | 19,303 | |
| Singapore (SG) | 37.0% | 3.3% | 95.1% | 25,559 | |
| Malaysia (MY) | 37.0% | 17.4% | 73.5% | 15,269 | |
| Colombia (CO) | 36.3% | 16.4% | 82.6% | 11,739 | |
| Vietnam (VN) | 32.1% | 20.1% | 77.9% | 20,568 | |
| Norway (NO) | 32.0% | 6.0% | 90.0% | 20,282 | |
| Armenia (AM) | 30.9% | 13.8% | 81.1% | 5,388 | |
| Australia (AU) | 30.8% | 0.5% | 91.5% | 33,198 | |
| Brazil (BR) | 29.6% | 12.2% | 86.3% | 18,600 |
received a provisional refusal refusal still stands
Why almost everyone is refused
A "provisional refusal" in Madrid vocabulary is what a U.S. filer calls an office action: the examining attorney has found something to object to, and the applicant has a fixed period to respond. It is not a decision. In the U.S. system nearly every application receives one, and Madrid-based applications (filed under §66(a) of the Lanham Act) hit an extra set of tripwires:
- Identification of goods and services. Most offices accept the broad Nice class headings an international registration is drafted with. The USPTO does not: it requires goods to be described with commercial specificity, and refuses "clothing" or "software" until narrowed. This single objection accounts for the bulk of §66(a) refusals, and it is why the rate is near-uniform across origin countries and classes (below).
- Disclaimers. Descriptive words inside a mark must be disclaimed apart from the mark as a whole, a U.S. formality most home registrations never needed.
- Translation and transliteration statements for any non-English wording.
- Likelihood of confusion (§2(d)) and descriptiveness (§2(e)). The substantive grounds: present, but a minority of first refusals.
A U.S. provisional refusal is the start of prosecution, not the end of it. Roughly 36 in 37 provisional refusals are overcome, usually by amending the goods and adding a disclaimer.
It is the system, not the applicant
If the refusal rate were about applicant quality, it would vary by where the mark comes from. It barely does. Whether the international registration originated in the EU, the UK, Japan, China or Switzerland, the provisional rate stays between 89.3% and 98.5%, and the share still refused between 2.0% and 2.9%.
| Origin office | Provisional refusal | Provisional | Still refused | Protected | Designations |
|---|---|---|---|---|---|
| European Union (EM) | 97.5% | 2.1% | 83.0% | 10,113 | |
| United Kingdom (GB) | 97.2% | 2.0% | 80.5% | 4,064 | |
| Japan (JP) | 89.3% | 2.6% | 85.9% | 3,163 | |
| Germany (DE) | 96.1% | 2.7% | 82.0% | 3,117 | |
| China (CN) | 94.0% | 2.9% | 70.5% | 3,081 | |
| France (FR) | 96.9% | 2.3% | 81.1% | 2,893 | |
| Switzerland (CH) | 92.9% | 2.3% | 83.8% | 2,714 | |
| Italy (IT) | 94.8% | 2.1% | 81.3% | 2,375 | |
| Australia (AU) | 98.5% | 2.5% | 77.7% | 2,224 | |
| Benelux (BX) | 98.2% | 2.1% | 84.5% | 1,731 | |
| South Korea (KR) | 94.0% | 2.0% | 82.1% | 1,646 |
The same flatness shows up by Nice class. The most-designated classes, software and electronics (9), IT services (42) and advertising and business services (35), all sit within a point or two of each other:
| Nice class | Provisional refusal | Provisional | Still refused | Protected | Designations |
|---|---|---|---|---|---|
| Class 9 Computers & Electronics | 98.0% | 2.5% | 81.0% | 15,221 | |
| Class 42 Computer & Scientific | 98.9% | 2.7% | 81.2% | 10,940 | |
| Class 35 Advertising & Business | 98.9% | 3.2% | 78.9% | 10,237 | |
| Class 41 Education & Entertainment | 98.8% | 2.7% | 79.4% | 6,502 | |
| Class 25 Clothing | 97.5% | 2.9% | 79.2% | 5,096 | |
| Class 5 Pharmaceuticals | 93.2% | 2.5% | 82.0% | 4,781 | |
| Class 3 Cosmetics & Cleaning | 96.1% | 2.9% | 79.9% | 4,460 | |
| Class 7 Machinery | 97.2% | 2.7% | 83.3% | 4,180 | |
| Class 10 Medical Apparatus | 96.5% | 1.8% | 84.2% | 3,382 | |
| Class 37 Building & Construction | 98.8% | 2.7% | 83.4% | 3,181 | |
| Class 1 Chemicals | 95.6% | 2.3% | 86.2% | 3,145 | |
| Class 16 Paper Goods | 98.5% | 2.7% | 81.7% | 3,088 |
Mark type makes a small difference: word marks are provisionally refused 94.2% of the time, combined word-and-design marks 97.9%, purely figurative marks 98.0%. Design elements add drawing-description and colour-claim formalities to the list of things an examiner can object to.
What "still refused" means, and the timing trap
A designation counts as still refused when its most recent examination outcome is a total or partial refusal with no later grant of protection. For the 2019–2022 cohort that is 2.6%, mostly marks whose owners never responded, or lost on likelihood of confusion or descriptiveness. Younger designations look far worse, because they simply have not resolved yet:
| Registration year | Designations | Provisional | Still refused | Protected | Status |
|---|---|---|---|---|---|
| 2018 | 9,787 | 95.0% | 2.1% | 82.2% | older: post-grant events creeping in |
| 2019 | 10,315 | 95.8% | 1.7% | 82.7% | cohort |
| 2020 | 10,865 | 95.6% | 1.0% | 82.8% | cohort |
| 2021 | 10,444 | 96.0% | 1.2% | 80.6% | cohort |
| 2022 | 12,523 | 96.1% | 5.8% | 75.4% | cohort |
| 2023 | 18,618 | 96.2% | 6.1% | 69.6% | resolving |
| 2024 | 24,910 | 94.5% | 3.9% | 58.1% | resolving |
| 2025 | 23,819 | 93.3% | 66.8% | 10.6% | still in examination |
| 2026 | 12,332 | 35.0% | 34.0% | 0.0% | still in examination |
What this means if you are filing
- Budget for the office action, not against it. Treat a U.S. response as a line item of the Madrid extension. A U.S.-licensed attorney is required to respond, and the response is routine when the objections are goods and disclaimers.
- Draft the goods for the USPTO before you file. Narrow, commercial descriptions in the international application avoid the most common refusal, even though other offices would have accepted the class heading.
- Do not read "provisional refusal" as a verdict. The likelihood of the designation being refused at the end of the road is closer to 2.6% than 95.9%. Check the substantive risk (an existing confusingly similar mark) with a clearance search before you file, because that is the refusal you cannot amend your way out of.
Comparing offices? The international refusal-rate pages break every major office down by class and origin, and the companion article looks at the reverse direction: where U.S. trademark owners get refused abroad.
Methodology
- Source. WIPO Madrid Monitor daily update files (ST.96 XML), ingested since July 2025: every international registration with any recorded activity in that window, 96,915 registrations carrying 974,457 designations in the cohort below. Each record carries the full event history of every designation, so outcomes are read from the events themselves, not inferred.
- Cohort. International registrations registered 2019–2022. Older years are increasingly dominated by post-grant events (U.S. §71 cancellations, renewals); newer years have not resolved.
- Provisional refusal = the office recorded a total or partial provisional refusal at any point. Still refused = the latest examination outcome is a total or partial refusal, with no later grant of protection; refusals overcome, reversed on review, or followed by a "possible opposition" or "intermediate status" notice are counted as overcome. Post-grant invalidations are excluded from refusals. Protected = a grant of protection (Rule 18ter or equivalent) is on record.
- Known bias. The feed is an activity delta, so registrations with no event since July 2025 are absent. Within the cohort this under-samples quiet, trouble-free designations slightly, which would make true refusal rates marginally lower than shown. Offices differ in how completely they report to WIPO; a few (e.g. Jersey, some African offices) report almost no examination events and show near-zero rates for that reason.
- Refresh. Figures recompute daily after ingest; the page shows the date of the data it renders. Percentages are rounded to one decimal.
International data © WIPO, re-used under CC BY 4.0; outcome normalisation is ours and WIPO bears no responsibility for it. Full details on the methodology and data sources pages.
Download the data
- offices.csv: all designated offices with ≥5,000 designations in the cohort
- us-by-origin.csv: U.S. designations by office of origin
- us-by-class.csv: U.S. designations by Nice class (top 20)
- us-by-year.csv: U.S. designations by registration year, 2018 onward
CSV, UTF-8, one row per office/class/year with counts and percentages. Free to reuse with a link to this page.