In FY2025, 2,059 people sat Japan's National Unified Examination for Sign Language Interpreters. 171 passed.
2,059 sat → 171 passed
In short: in the 25 years since the exam began in 2001, no other year has had a single-digit pass rate. The previous low was 11.77% in FY2014. At 8.31%, FY2025 sits more than three points below even that.
And in the same year the number of candidates hit an all-time high. The largest field the exam has ever drawn produced its lowest pass rate.
(the only single-digit year in 25)
(a record, and the first year above 2,000)
(333 the year before)
→ Skip to what happened in FY2025 / → Jump to the 25-year table
Incidentally: “the exam is run prefecture by prefecture, so nobody compiles a national pass rate.” Plenty of pages say this — and, to my embarrassment, so did another article on this very site, written by me.
But on looking properly, I was able to confirm the figures in official documents. The national totals are published every single year, and a table reaching back to 2001 is public as well. This article shows exactly which document each figure comes from.
This article lays out all 25 years — applicants, candidates who sat, passes, and pass rates — from the primary sources. I also say plainly where those sources disagree with the tables circulating online.
A note for readers outside Japan. Japan has two separate credentials. The National Unified Examination (統一試験) is what this article is about: pass it and you register as a working interpreter with your prefecture — a 手話通訳者 (shuwa tsūyakusha), called a prefecturally registered interpreter below. Separately, the national certification examination (手話通訳技能認定試験) is a Ministry-recognised credential taken at four sites nationwide, and its holders are 手話通訳士 (shuwa tsūyakushi), called nationally certified interpreters below. The two Japanese words differ by a single character — 者 (sha, “person who does”) and 士 (shi, “licensed professional”). They are different exams with different candidate pools.
Where the national figures actually live
The exam papers are written by the national sign language training centre (全国手話研修センター), a social welfare corporation in Kyoto. The exam itself is administered by deaf-related organisations in each prefecture and government-designated city, and each of those bodies announces its own results. That structure is why people assume no one aggregates the numbers.
But a social welfare corporation in Japan is legally required to publish an annual business report. The centre posts its annual report (事業報告書) as a PDF on its own site each June — and the exam figures are written out in the body text.
Exactly where the numbers are
• In the main text, under “Part 2 / Human resource development / Section 2 Independent projects / 1. National Unified Examination”, each report states that year's applicants, candidates who sat, passes and pass rate, with a comparison to the previous year.
• Reports up to FY2021 also carry an appendix table, “Passes by region”, which contains candidates, passes and pass rates for every year back to 2001. That table is the source for 2001–2021 here.
• The PDFs published from FY2022 onward contain the main text only, with no appendix. So prefecture-level figures stop at FY2021 (more on this below).
[Correction notice] My own guide article, “The National Unified Examination for Sign Language Interpreters — a complete guide”, previously stated that the pass rate is not officially made public. Now that I have confirmed the figures in the official documents, that article has been corrected too.
The exam itself began in 2001, first administered by the centre's founding preparatory office — the annual reports say so directly. That is why the table below starts there.
25 years of national data
National totals for FY2001–FY2025. Applicant counts are shown only for the years in which they were published.
| Fiscal year | Applied | Sat | Passed | Pass rate |
|---|---|---|---|---|
| 2001 | — | 387 | 109 | 28.17% |
| 2002 | — | 618 | 220 | 35.60% |
| 2003 | — | 621 | 156 | 25.12% |
| 2004 | — | 839 | 227 | 27.06% |
| 2005 | — | 894 | 280 | 31.32% |
| 2006 | — | 1,093 | 332 | 30.38% |
| 2007 | — | 1,455 | 251 | 17.25% |
| 2008 | — | 1,420 | 403 | 28.38% |
| 2009 | — | 1,495 | 177 | 11.84% |
| 2010 | — | 1,584 | 428 | 27.02% |
| 2011 | — | 1,507 | 486 | 32.25% |
| 2012 | — | 1,471 | 242 | 16.45% |
| 2013 | — | 1,554 | 277 | 17.82% |
| 2014 | — | 1,504 | 177 | 11.77% |
| 2015 | — | 1,618 | 292 | 18.05% |
| 2016 | — | 1,713 | 247 | 14.42% |
| 2017 | — | 1,800 | 232 | 12.89% |
| 2018 | 1,875 | 1,762 | 316 | 17.93% |
| 2019 | 2,014 | 1,881 | 377 | 20.04% |
| 2020 | 1,457 | 1,348 | 282 | 20.92% |
| 2021 | 1,531 | 1,444 | 295 | 20.43% |
| 2022 | 1,667 | 1,535 | 281 | 18.31% |
| 2023 | 1,777 | 1,690 | 256 | 15.15% |
| 2024 | 1,962 | 1,852 | 333 | 17.98% |
| 2025 | 2,181 | 2,059 | 171 | 8.31% |
| Total (25 years) | — | 35,144 | 6,847 | 19.48% |
The shaded rows are the lowest pass rate in the 25 years (FY2025, 8.31%) and the highest (FY2002, 35.60%).
Roughly 35,000 people have sat the exam in 25 years and about 6,800 have passed — about one in five.
The trend, in two charts
The chart scrolls sideways on a narrow screen. “View larger” opens it full-size.
The first thing the chart shows is that the first six years and everything after look like two different exams. FY2001–2006 sat in the 25–35% band, with 400 to 1,100 candidates a year.
Then in FY2007 the candidate count jumped to 1,455 and the pass rate fell to 17.25%. In the nineteen years since, only six have cleared 20%; the rest sit in the teens. As the pool grew, the whole level of the pass rate stepped down.
The chart scrolls sideways on a narrow screen. “View larger” opens it full-size.
Counted in people rather than percentages, the peak was FY2011 with 486 passes. The line swings a lot but generally moves between 200 and 400. Against that, FY2025's 171 passes is the third-lowest total on record, behind FY2001 (109) and FY2003 (156). But those were years in which only 387 and 621 people sat the exam. 171 passes out of a field of 2,000 has no parallel in the series.
A closer look at the last eight years
From FY2018 the applicant count is published too, so we can also see how many people applied and then did not sit.
| Fiscal year | Sites | Applied | Sat | No-show rate | Passed | Pass rate |
|---|---|---|---|---|---|---|
| 2018 | 50 bodies (46 prefectures + 4 designated cities) | 1,875 | 1,762 | 6.0% | 316 | 17.93% |
| 2019 | 50 bodies (46 prefectures + 4 designated cities) | 2,014 | 1,881 | 6.6% | 377 | 20.04% |
| 2020 | 49 bodies (44 prefectures + 5 designated cities) | 1,457 | 1,348 | 7.5% | 282 | 20.92% |
| 2021 | 51 bodies | 1,531 | 1,444 | 5.7% | 295 | 20.43% |
| 2022 | 51 bodies (46 prefectures + 5 designated cities) | 1,667 | 1,535 | 7.9% | 281 | 18.31% |
| 2023 | 51 bodies (46 prefectures) | 1,777 | 1,690 | 4.9% | 256 | 15.15% |
| 2024 | 51 bodies | 1,962 | 1,852 | 5.6% | 333 | 17.98% |
| 2025 | 52 bodies (all 47 prefectures) | 2,181 | 2,059 | 5.6% | 171 | 8.31% |
The no-show rate stays roughly between 5% and 8% throughout. So the movement in the pass rate cannot be explained by a sudden change in who turns up.
Candidate numbers moved a lot more: down to 1,348 in the pandemic year FY2020, then climbing for five straight years to a record 2,059 in FY2025. The pass rate went the other way, from 20.92% in FY2020 to 8.31% in FY2025.
What happened in FY2025
Four records moved in the same year.
1. Applicants: 2,181 — an all-time high
2. Candidates who sat: 2,059 — an all-time high, and the first time above 2,000
3. Pass rate: 8.31% — the only single-digit year in 25
4. 52 administering bodies, covering every prefecture — the first time in the published record, and a goal the centre had been stating in its reports for years
Year on year, 207 more people sat the exam, and 162 fewer passed it: 333 down to 171.
A single-digit pass rate has not happened in any other year on record. The previous low was 11.77% in FY2014, followed by 11.84% in FY2009. At 8.31%, FY2025 is more than three points below either.
On the cause, the centre itself wrote in its FY2023 report that the decline is likely linked to the pandemic: fewer opportunities to communicate in sign language, and not enough group study.
There is also a change on the exam side. From FY2020 the practical test became situational interpreting only. The “summarising in sign language” component that existed before was folded into situational interpreting and into the written Japanese-language paper, according to the annual report. The content of the exam itself shifted in 2020, and that is worth holding in mind.
None of this proves the exam “got harder”. A pass rate moves with who is sitting it (a wave of newer candidates pushes it down) as much as with how it is marked. What we can say cleanly is the sequence of facts: the year the exam finally reached every prefecture and drew its largest-ever field is also the year its pass rate hit its lowest point. That is all I would ask you to take from it.
For reference: alongside the certification exam
The unified exam is often compared with the Ministry-recognised national certification examination for sign language interpreters (手話通訳技能認定試験), run by a separate corporation, which publishes its own results back to its first sitting. Lined up by fiscal year:
| Fiscal year | Unified exam pass rate | Shuwa-tsuyakushi exam (no.) | Shi exam pass rate |
|---|---|---|---|
| 2019 | 20.04% | No. 30 | 9.8% |
| 2020 | 20.92% | No. 31 | 11.0% |
| 2021 | 20.43% | No. 32 | 9.6% |
| 2022 | 18.31% | No. 33 | 13.3% |
| 2023 | 15.15% | No. 34 | 12.2% |
| 2024 | 17.98% | No. 35 | 5.5% |
| 2025 | 8.31% | No. 36 | 11.2% |
Shaded rows are years in which the certification exam had the higher pass rate of the two.
In FY2025 the unified exam (8.31%) came in below the certification exam (11.2%). That is not unprecedented: the same reversal happened in FY2007, FY2009, FY2010, FY2012 and FY2014. The belief that the certification exam is always the harder of the two did not hold up in the numbers.
These are not directly comparable. Eligibility and candidate pools differ. The certification exam is held at four sites (Saitama, Tokyo, Osaka, Fukuoka) and draws many people who already interpret professionally. The unified exam is taken locally by people who have completed an interpreter training course.
How to read these numbers
The denominator is candidates who sat, not applicants
The published pass rate is passes divided by candidates who sat. Dividing by applicants gives a figure a few points lower. For all 25 years in the table above, passes divided by candidates equals the published rate.
“National” does not always mean all 47 prefectures
In FY2020, Tokyo and Shizuoka cancelled because of the pandemic, leaving 49 administering bodies. Osaka Prefecture ran its own separate registration exam for many years and did not use the unified exam (the designated city of Sakai, within Osaka, did). So the national pass rate is the total of whoever administered it that year, and the coverage varies across the series.
Some tables online disagree with the source
Pass-rate tables for this exam circulate on several sites without citations. Most years match the annual reports — but FY2020 does not. The widely-copied table gives 266 passes and 19.73%; the annual report's appendix gives 282 passes and 20.92%. The prefecture-by-prefecture breakdown in that same appendix sums to 282, so the internal figures agree with the total. This article uses the annual report.
Prefecture-level data stops at FY2021
Annual reports through FY2021 include a table of applicants, no-shows, candidates and passes for every prefecture and designated city. That is the last year for which a prefecture-level pass rate can be calculated. The PDFs published from FY2022 onward have no appendix, so the last four years of prefecture data are not public.
Not everyone is allowed to sit it
The candidate count is not a count of people who want to become interpreters. The training centre’s handbook names only two eligible groups: people who have completed an interpreter training course, and people with equivalent knowledge and skill. Without a place on a training course, you cannot reach the starting line.
Each administering body then narrows it further. The Kanagawa Deaf Welfare Centre’s FY2026 notice, for example, requires that you intend to work as an interpreter in the prefecture and to register with the prefecture and your municipality after qualifying, and then limits entry to either people finishing the centre’s own training course that year, or people aged 18 or over who have completed a training course (in an area that uses the unified exam) or graduated from a specialist school such as the National Rehabilitation Center, and who live in Kanagawa outside Yokohama and Kawasaki (proof required).
So the ceiling on candidate numbers is set by how many people finish a training course locally. That narrow entrance is part of why the pass rate cannot be read as a measure of difficulty alone.
The prefecture-level numbers are a large topic in their own right, so they have an article of their own. Hyogo, not Tokyo, leads on cumulative passes — and per 100,000 residents the ranking almost entirely inverts.
Hyogo, Not Tokyo, Tops the List — prefecture data, 2001–2021What interpreters are actually paid
Passing this exam lets you register as an interpreter with your prefecture, and each assignment carries a fee set by the local government. The same two-hour assignment pays differently depending on where you are. I have mapped the real per-municipality amounts, sourced from each government's own regulations.
See the interpreter fee map →Sources
Every figure in this article comes from the following primary sources.
- National sign language training centre (全国手話研修センター), FY2025 Annual Report
FY2025 annual report (PDF, Japanese) — FY2025 applicants, candidates, passes, pass rate - Same, FY2024 / FY2023 / FY2022 Annual Reports — FY2022–FY2024 figures
- Same, FY2021 Annual Report, appendices 2 and 2-2
FY2021 annual report (PDF, Japanese) — candidates, passes and pass rates for 2001–2021, and passes by prefecture - Same, FY2020 Annual Report, appendix 4-1
FY2020 annual report (PDF, Japanese) — FY2020 prefecture breakdown; the change to the practical test - Same, FY2019 Annual Report — FY2018 and FY2019 applicant counts
- National Sign Language Training Center, Handbook for the FY2026 National Unified Examination (eligibility and exam subjects; distributed by each administering body)
- Kanagawa Deaf Welfare Centre, Notice for the FY2026 National Unified Examination (an example of how one body sets eligibility)
Kanagawa Deaf Welfare Centre notices (Japanese) — accessed 21 August 2026; it appears to be posted only during the application period, so the link may not resolve later. - Centre disclosure page
https://www.com-sagano.com/center/ - Jyoubun Center (聴力障害者情報文化センター), “Applicants, candidates and passes for the national certification examination, 1st–36th sittings”
Examination results data
Note: the reports from FY2022 onward are scanned PDFs, so those figures were read visually from the page and transcribed. If you spot an error, please tell me through the contact form and I will correct the article.
Preparing for the written paper
Last year’s pass rate was the lowest on record, but the exam requires clearing both a written and a practical paper, and the written one has a published scope — which makes it the part where preparation shows most clearly.
Shunsoku Master: Sign-Language Interpreter Exam (on Google Play)
A study app focused on the written exam (basic knowledge and Japanese language). I spent 12 years teaching at a special-needs school for the Deaf, and wrote the practice questions to follow the scope the exam publishes.
717 multiple-choice questions (361 on terminology, 356 on Japanese), a 532-term glossary, 55 study stages and 10 full mock papers. You can narrow practice to a single field and re-drill only what you got wrong.
Free access covers term stages 1–3, Japanese stages 1–3, set 1 of every field, mock exam 1, and the full 532-term glossary. A premium plan unlocks all stages and sets, all 10 mock exams, and random practice, and removes the ads. The questions are not past papers — they are written to match the published scope. The app is in Japanese only.
Google Play and the Google Play logo are trademarks of Google LLC.
Conclusion
The national pass rate for Japan's unified sign language interpreter exam is published, every year, in the training centre's annual report.
Set out across 25 years, FY2025's 8.31% is the only single-digit year on record, well below the previous low of 11.77% in FY2014. The same year saw the largest candidate field ever recorded and, as far as the published reports show, the first sitting held in every prefecture.
That is one reading, not the reading.
Pass rates move with the composition of the candidate pool and with changes to the exam itself, and there is regional and individual variation behind every figure. My aim here is to put the sequence of facts on the table rather than to declare the exam harder.
If you know of an error, or of another source, please get in touch through the contact form and I will update the article.