Global Ultrasonic Rail Testing Standards and Regulations

Ultrasonic rail testing standards come in two kinds. Technical standards such as EN 16729-1 and ISO 5735-1:2024 define how in-track testing is done. Regulations set how often it must be done: in the US, 49 CFR 213.237 requires Class 4–5 track to be tested every 370 days or 30 MGT, whichever comes first.

That split is the most useful thing to know when comparing countries. A European or ISO standard tells you what a valid ultrasonic test system and procedure look like, but it doesn't set a test date. A national regulation, or a network owner's own standard, sets the interval and the action when a defect is found. This guide walks through each layer: Europe and ISO, the UK, the UIC, Australia, the US and Canada. It ends with a comparison table and a worked example for the same 25 MGT line under different regimes.

Contents: Technical standards vs regulations · EN 16729 · ISO 5735-1 · EN 13674 · UK and UIC · Australia · US · Canada · Comparison table · Worked example · Personnel · Limits · FAQ

What is the difference between a rail testing standard and a regulation?

A standard describes a method; a regulation makes a duty enforceable. A standard such as EN 16729-1 is voluntary until a law, a contract or a network owner's rulebook calls it up. A regulation such as 49 CFR 213.237 applies by force of law to the track owners it covers.

The two layers answer different questions:

Key takeaway: when a contract says "test to EN 16729-1", it fixes the method, not the interval. The interval has to come from somewhere else.

What does EN 16729 cover?

EN 16729 is the European (CEN) series for non-destructive testing of rails in track, and Part 1 is the ultrasonic part. EN 16729-1:2016, Railway applications – Infrastructure – Non-destructive testing on rails in track – Part 1: Requirements for ultrasonic inspection and evaluation principles, was published in July 2016 and confirmed in December 2021.

EN 16729-1 covers rails installed in track, tested with dedicated test vehicles or manually propelled devices. Its stated aim is comparable results on the location, type and size of discontinuities. It does not cover vehicle acceptance or testing in rail production plants, and its scope says it "applies only to rail profiles meeting the requirements of EN 13674-1". It also gives no guidance on managing the results, which is the owner's job.

BS EN 16729-1 is the UK adoption of the same European text, not a separate British standard. The "BS" prefix means BSI has published the EN unchanged as a British Standard; other countries do the same as DIN EN, NF EN or SS-EN. The original version of this article treated BS EN 16729-1 as the standard "EU operators follow", which mixed up the two. The European number is EN 16729-1.

The series now has five published parts, verified from national catalog listings:

Two scope limits matter in practice. EN 16729-3 "does not specify the assessment criteria of rail defects and the derived actions", so defect disposition stays with the infrastructure manager. EN 16729-5 doesn't set how many welds to test and doesn't cover welding procedure approval. A revision of Part 1 (prEN 16729-1) reached the CEN enquiry stage in 2026, so check the edition named in your contract.

Key takeaway: EN 16729 sets how rail is tested and who may test it, but never how often.

What does ISO 5735-1:2024 add?

ISO 5735-1:2024 is the international (ISO) standard for ultrasonic testing of rails in track. It was published on March 14, 2024 as Railway infrastructure – Non-destructive testing on rails in track – Part 1: Requirements for ultrasonic testing and evaluation principles.

Its scope, read on a national catalog listing, is close to EN 16729-1. It sets requirements for test principles and systems so that results are comparable on location, type and size of discontinuities. It covers testing by vehicles or manually propelled devices, "treats welded joints as plain rail", and "only applies to flat bottom (Vignole) railway rail profile 43 kg/m and above". Like the EN, it sets no test interval.

The original article described ISO 5735-1 as the "fundamental requirements" standard that defines "evaluation criteria … for categorizing defects and determining rail serviceability". That overstated it. ISO 5735-1 is about test principles and systems; deciding whether a rail stays in service is left to the owner's rules. The full text is paywalled, so this article cites only the published title and scope.

Key takeaway: ISO 5735-1 is the reference to cite outside Europe when you want a recognized test method, and it reaches lighter rail (43 kg/m and up) than the EN 13674-1 profiles that bound EN 16729-1.

Where does EN 13674 fit?

EN 13674 is a product standard for the rail itself, not an inspection standard. EN 13674-1 specifies Vignole (flat-bottom) railway rails of 46 kg/m and heavier for conventional and high-speed track, including the steel grades and profiles. EN 13674-2 covers switch and crossing rails used with them.

EN 13674 matters to in-track UT for one reason: it sets the scope boundary. EN 16729-1, -3 and -5 apply only to rail profiles meeting EN 13674-1, and EN 16729-4 refers to EN 13674-1 and -2. The Australian equivalent of a rail product standard is AS 1085.1:2019, Railway track material – Part 1: Steel rails, developed by RISSB and published by Standards Australia. It covers as-rolled and hardened rails from continuously cast blooms and excludes rails harder than 350 HB nominal. In North America, rail is bought to AREMA and ASTM specifications; see our AREMA vs FRA vs ASTM explainer for how those bodies divide the work.

Key takeaway: a product standard tells you what rail was supplied; it doesn't tell you how to test it once it's in track.

How do the UK and the UIC fit in?

In the UK, BS EN 16729-1 gives the method and Network Rail's own company standards set the inspection regime on its network. Network Rail's module NR/L2/TRK/001/mod06, Inspection and Maintenance of Permanent Way: Visual Inspection, Ultrasonic and Eddy Current Testing of Rails, is listed by a standards distributor as current from December 2020. The text isn't public, so this article doesn't quote its intervals. Treat any figure you see attributed to it secondhand with caution.

The UIC (International Union of Railways) publishes recommendations, not law. Its rail defect catalog, formerly UIC Leaflet 712 and now IRS 70712 Rail Defects (first IRS edition 2018), gives each defect a number and a description sheet. Equipment makers and some infrastructure managers use those numbers in reports. A companion document on rail maintenance and defect treatment, formerly UIC 725 and now IRS 71725, sits beside it. Both are hedged here because only their titles and descriptions were checked, not their full text. A railway adopts them by choice or by contract.

Key takeaway: in Europe, the method is standardized, while the interval and the defect response sit in each infrastructure manager's rulebook and safety management system. The EU Railway Safety Directive, (EU) 2016/798, requires infrastructure managers to run a safety management system, but EN 16729 itself doesn't make testing frequency a legal number.

How does Australia set rail testing intervals?

Australia's main interstate network manager, ARTC, sets ultrasonic testing intervals in its own engineering standards. ARTC's specification ETA-01-05, Non-Destructive Testing of Rail (version 1.0, reviewed March 2024), says the test frequency for each line section "shall be published in the Technical Maintenance Plan", and ties the frequency to a performance target:

"Testing should be undertaken at a frequency that supports maintaining the combined rail and weld breaks at or below 0.02 breaks per km per year." (ARTC ETA-01-05, section 3.1.3)

That break count includes only full breaks that continuous car testing could feasibly have found, and it excludes insulated rail joints. ARTC's Civil Technical Maintenance Plan (ETP-00-03, version 1.5, July 2025) sets the default: "Where not otherwise specified … the frequency will be in accordance with the CoP requiring a test at least every 15 MGT." Busy heavy-haul coal lines in the Hunter Valley carry much shorter calendar intervals in the same plan, some as short as 31 days. Changing a frequency requires ARTC's change management process, a risk assessment and notice to regulators.

For personnel, ETA-01-05 section 2.1 requires that rail NDT be done only by people assessed as competent under ARTC's written practice ETP-00-01, and it lists AS ISO 9712 among its references. The 2024 revision removed an earlier reference to an ASNT document.

Key takeaway: ARTC's model looks like FRA's in one way (a performance target) and unlike it in another (a tonnage interval set by the owner, not by regulation). Other Australian networks have their own standards, so check the network you're working on.

What do US regulations require?

In the US, FRA's Track Safety Standards make internal rail testing a legal duty on the general railroad system. The core sections are short to list; our FRA track safety standards guide and 49 CFR 213.237 guide cover them in depth.

The original article said "recent updates even allow continuous ultrasonic rail testing" without a citation. The update is 49 CFR 213.240, added by FRA's rule of October 7, 2020 (85 FR 63362). It's an option, not a requirement; see continuous vs stop-and-verify rail testing for how the two test types differ. Note also that Part 213 never uses the word "ultrasonic": the rule requires "internal rail inspections" with equipment that finds defects between joint bars (213.237(f)), and UT is the industry's way of meeting it.

Two US points often confuse international readers:

• AREMA isn't a regulator. Its Manual for Railway Engineering is published "as a guide to railways in establishing their individual policies and practices". FRA's Part 213 is the law.
• Isolated rail transit isn't under Part 213. Rapid transit not connected to the general system falls under FTA Parts 671–674 and a state safety oversight agency, and those federal rules set no rail flaw test interval. APTA's voluntary RT-FS-S-002-02 says mainline ultrasonic testing "shall be performed at least once per year", which applies only where an agency adopts it.

Key takeaway: the US is the only jurisdiction here where both the interval and the operator qualification are federal law.

What do Canada's Track Safety Rules require?

Canada sets rail flaw testing frequency by track class and annual tonnage in Transport Canada's Rules Respecting Track Safety (TSR). The rules apply to federally regulated railway companies operating on standard gauge track (sections 3.1 and 4.1). The version checked here is dated May 31, 2022 and in force from May 31, 2023; check Transport Canada for later amendments.

TSR Part II, Subpart F, section 5 starts with a definition close to FRA's:

"A Rail Flaw Inspection is a continuous search for internal rail defects." (TSR Part II, Subpart F, 5.1)

Section 5.2 then sets a minimum number of tests a year:

Each frequency also carries a maximum gap and, for the higher frequencies, a winter test (November 15 to March 31):

Other TSR provisions that differ from FRA:

• New rail credit (5.2(a)). If new rail is ultrasonically inspected over its entire length within 6 months of installation and all defects are removed, the next search isn't due until 100 MGT or three years, whichever comes first.
• Heavy-axle rules (5.3, 5.4). Class 2 track carrying 3 to 5 MGT with rail under 100 lb and cars of 263,000 lb or more must be tested at least annually. Class 4 or 5 track with 100 lb rail and cars of 286,000 lb or more must be tested twice annually.
• Sidings and crossovers (5.5). Rail where track speed is 25 mph or more must be inspected annually.
• Missed segments (5.8). Where a valid search can't be made, the railway must, before the time or tonnage limit expires, re-search, reduce the class of track, or remove the rail. FRA's equivalent option is a 25 mph limit rather than a class reduction.

Equipment must detect defects between joint bars (5.6), and defective rail must be marked and highly visible (5.7), both parallel to 213.237(e) and (f).

Key takeaway: Canada's table is calendar-count based within tonnage bands, so busier lines get more tests a year, not a tonnage limit between tests.

How do ultrasonic rail testing standards compare by country?

Only the US and Canada set a rail flaw test interval in national regulation; Europe and ISO standardize the method and leave the interval to the network owner. The table below summarizes the documents covered above.

Key takeaway: if you're writing a multinational specification, name the method standard and the interval source separately.

Worked example: one 25 MGT line under four regimes

For a Class 4-equivalent main line carrying 25 MGT a year, FRA requires about one test a year, Canada three, and ARTC's 15 MGT default about 1.7. The inputs are the same for each regime: 25 MGT a year, spread evenly, on a line rated Class 4 under the US and Canadian systems. The class systems are similar but not identical, so treat this as an illustration, not a legal equivalence.

Three things stand out:

1. The same traffic can produce a threefold difference in tests. Canada's three tests a year compare with FRA's one. FRA compensates with a performance check: if service failures exceed 0.10 per mile per year for two years, the owner must move to 10 MGT testing or reduce the class.
2. FRA's interval is a ceiling, not a target. An owner may test more often than 213.237 requires, and the legal minimum isn't necessarily the right interval for older rail or a segment with a failure history.
3. The European row can't be filled with a single number. Any figure presented as "the European interval" for in-track UT should be traced to a named infrastructure manager's standard. EN 16729-1 doesn't contain one.

For the US calculation in more depth, including the "whichever is longer" trap for other Class 3 track, see how often must rail be tested.

Who is qualified to test: ISO 9712, EN 16729-4 and SNT-TC-1A?

Personnel rules split into two models: third-party certification (ISO 9712) and employer certification (SNT-TC-1A), with rail-specific layers on top. Which one applies depends on the jurisdiction and the contract.

EN 4179 sometimes appears in rail discussions, but it is an aerospace-series standard for NDT personnel in aerospace manufacturing and maintenance (the current edition is EN 4179:2026). It has no role in rail testing. If a rail specification cites it, the citation is probably wrong.

Under SNT-TC-1A, ASNT doesn't certify the technician; the employer does. FRA doesn't certify anyone either: 213.238 requires the employer to qualify and authorize each operator on each type of equipment. A European tender is more likely to ask for ISO 9712 certification plus EN 16729-4 rail qualification, while a US tender asks for 213.238 records and often SNT-TC-1A certification. Our SNT-TC-1A requirements guide covers the employer model in detail.

Key takeaway: a general NDT certificate doesn't on its own prove someone can run rail flaw equipment. Both the US rule and EN 16729-4 add a rail-specific layer.

What these standards don't cover

No standard or regulation in this guide guarantees that every flaw is found. They set method, frequency and competence, and each has gaps an engineer should state in a specification:

• Grooved rail. Neither EN 16729-1 nor ISO 5735-1 covers grooved (tram) rail, so street-running systems need their own procedures and reference pieces made from the same rail section.
• Defect disposition. EN 16729-1 gives no guidance on managing results, and EN 16729-3 sets no assessment criteria. The action on a defect comes from the owner's rules or, in the US, 213.113.
• Surface masking. Shelling, head checks and other surface damage can block ultrasound from reaching a detail fracture beneath them. FRA's 2014 rule preamble notes the "limitations of current technology to detect internal rail flaws beneath surface conditions and in the base flange area".
• Rail base coverage. In-track UT reaches from the rail head only. Coverage of the base outside the area under the web is limited.
• Welds. ISO 5735-1 treats welded joints as plain rail. Detailed weld inspection in track needs EN 16729-5 or an owner procedure.
• Intervals outside the US and Canada. This guide verified ARTC's published standards for Australia only. Other networks, including those in Europe and the UK, set their own intervals, which aren't all public.

How EMA Industries applies this

EMA Industries performs ultrasonic rail flaw testing for transit agencies, short lines and industrial track owners. We're based in Southern California and mobilize nationwide. Most US work runs to 49 CFR 213.237 and 213.238, or to a transit agency's own program. When an owner's specification also cites EN 16729-1 or ISO 5735-1, the test procedure, reference rail and report can be mapped to both. To scope a testing program, see our ultrasonic rail testing services or contact us.

Related reading: 49 CFR 213.237 rail flaw detection · Continuous vs stop-and-verify rail testing · FRA track safety standards · AREMA vs FRA vs ASTM · How often must rail be tested?

Frequently asked questions

Is EN 16729-1 mandatory in Europe?

Not by itself. EN 16729-1 is a voluntary European standard. It becomes binding when an infrastructure manager's standard, a national rule or a contract calls it up, and the test interval still comes from the infrastructure manager.

What is the difference between EN 16729-1 and BS EN 16729-1?

They are the same text. EN 16729-1 is the European standard; BS EN 16729-1 is its UK adoption published by BSI, just as DIN EN 16729-1 is the German adoption.

How often must rail be ultrasonically tested in Canada?

It depends on track class and annual tonnage under section 5.2 of Transport Canada's Track Safety Rules. Class 4 track carrying more than 15 and up to 35 MGT a year needs three tests a year, no more than 250 days apart, with at least one in the winter period.

Does ISO 5735-1 set a test frequency?

ISO 5735-1:2024 sets requirements for ultrasonic test principles and systems, with no test interval. The frequency comes from the regulator or the network owner, for example 213.237 in the US.

Which personnel certification is accepted for rail UT internationally?

ISO 9712 third-party certification is common outside North America, and EN 16729-4 adds rail-specific qualification in Europe. In the US, FRA requires a 213.238 qualified operator, and contractors often certify UT staff to SNT-TC-1A under an employer written practice.

Sources

• EN 16729-1:2016 listing and scope, genorma.com; BS EN 16729-1 on BSI Knowledge
• EN 16729-2:2020 (eddy current), iTeh catalog
• EN 16729-3:2018 scope, genorma.com
• SS-EN 16729-4:2019 (personnel), Swedish Institute for Standards
• BS EN 16729-5:2023 (welds in track), BSI Knowledge
• ISO 5735-1:2024, ISO; scope listing, genorma.com
• AS 1085.1:2019 Steel rails, RISSB
• UIC Leaflets / IRS numbering correlation table, UIC
• ARTC ETA-01-05 Non-Destructive Testing of Rail (v1.0, 2024) and ETP-00-03 Civil Technical Maintenance Plan (v1.5, 2025)
• Transport Canada, Rules Respecting Track Safety, Subpart F (May 31, 2022 version); consolidated PDF (TC E-80)
• 49 CFR 213.237, 213.238, 213.240, 213.113 and 213.339, eCFR; text read on Cornell LII
• Track Safety Standards; Improving Rail Integrity, 79 FR 4234 (January 24, 2014), Federal Register
• Rail Integrity and Track Safety Standards, 85 FR 63362 (October 7, 2020), Federal Register
• ASNT standards (SNT-TC-1A, CP-189), ASNT
• EVS-EN 4179:2026, Estonian Centre for Standardisation
• APTA RT-FS-S-002-02 Rev. 1, APTA