Continuous vs Stop-and-Verify Rail Testing: Which Is Right?

Continuous vs stop-and-verify rail testing comes down to when each suspect location is checked. Stop-and-verify testing confirms indications at the rail within 4 hours, or immediately for suspected A, A2 and B defects (49 CFR 213.113(b)). Continuous testing under 49 CFR 213.240 keeps the vehicle moving and allows 84 hours, or 36 hours for suspected A, A2 and B defects.

Both methods satisfy the same internal rail inspection requirement in 49 CFR 213.237, and both end in the same remedial action table once a defect is confirmed. What changes is the workflow in between: who reads the data, where they read it, how long the owner may wait before a person with a hand probe stands over the suspect spot, and what paperwork goes to FRA each year.

This guide sets out the rule text, the four test processes FRA recognizes, how suspects are prioritized, what the evidence from the waiver years shows, a decision table and a worked example on a 150-mile Class 4 line.

Contents: Difference · 213.240 rules · Verification windows · Prioritization · Evidence · Equipment · Decision table · Worked example · Cost · Limits · FAQ

Continuous vs stop-and-verify rail testing: what is the difference?

The difference is when the indication is confirmed. In a stop-and-verify test the vehicle stops at each indication and an operator confirms it on the spot. In a continuous test the vehicle runs non-stop, the data is analyzed elsewhere, and verification crews follow later.

FRA described four rail flaw detection processes in the preamble to its October 7, 2020 Rail Integrity and Track Safety Standards final rule (85 FR 63362), which created 213.240. Each one needs a person to interpret the data:

Before 2020, the 4-hour verification limit in 213.113(b) made a true continuous test impractical. FRA said so plainly in the same preamble: the timeframe "has made it practically impossible for track owners to conduct continuous testing." Railroads ran continuous programs only under conditional waivers, starting with CSX in 2009, and the 2020 rule turned that waiver practice into an option any track owner can elect.

Key takeaway: the test vehicle and the probes can be the same; the regulatory regime and the verification workflow are what differ.

What does 49 CFR 213.240 require for continuous rail testing?

Section 213.240 is an elective procedure, and it comes with conditions. Under paragraph (a), a track owner "may elect" continuous testing to satisfy 213.237 (Class 1–5) or 213.339 (Class 6 and higher). An owner who does so "is exempt from the requirements of § 213.113(b); all other requirements of § 213.113 apply." Only the verification clock changes. The defect definitions, the remedial action table and the 213.113(a) duties still apply.

The rest of 49 CFR 213.240 (read on the Cornell LII copy of the eCFR, October 2026) sets these conditions:

The paragraph that matters most for anyone managing the defect list is (e)(4):

"A suspect location is not considered a defect under § 213.113(c) until it has been field-verified by a person qualified under § 213.238. After the suspect location is field-verified and determined to be a defect, the track owner must immediately perform all required remedial actions prescribed in § 213.113(a)." (49 CFR 213.240(e)(4))

The 213.238(b)(3) exception in paragraph (d)(1) is easy to miss. On a stop-and-verify car, the qualified operator interprets responses and acts on them. On a continuous car, interpretation moves to the analyst under (d)(2), so the vehicle operator doesn't need that authorization. The analyst's qualification record becomes a compliance document of its own.

Key takeaway: 213.240 trades a longer verification clock for written procedures, separate analyst qualification, pre-test designation and an annual report to FRA.

How do the 84-hour and 36-hour windows compare with the 4-hour rule?

Stop-and-verify works to hours; continuous works to days. Here is the comparison, from the rule text of 213.113(b) and 213.240(e):

Two details catch owners out.

The clocks start at different moments. Under 213.113(b), the duty runs from when the owner "learns" of an indication, so the contractor-to-owner notice path matters. Under 213.240(e), the 84 and 36 hours run "from indication of the suspect location", meaning the moment the car records it, not the moment an analyst reports it. Analysis time comes out of the window.

Switching type mid-run switches the rule. The FRA preamble says that if an owner changes from a continuous test to a stop-and-verify test, "all requirements of Sec. 213.113 will immediately apply" and suspects found in the stop-and-verify portion must be verified within 4 hours. The mileposts and times of the change have to be recorded under 213.240(c).

The 36 and 84 hours aren't response times for every defect on the railroad (a common error in drafts on this topic). They apply only to suspects from a properly designated continuous test. Indications from a stop-and-verify run, a portable test or a track inspector stay under the 4-hour rule.

Key takeaway: the 84-hour window starts when the car records the suspect, so analysis time is part of the budget.

How are suspect locations prioritized?

By the analyst's estimate of defect type and size, sorted into priority one, two and three. Paragraph (b)(3) of 213.240 requires suspects to be "categorized and prioritized according to their potential severity." The rule doesn't define the categories. The 2020 preamble describes how the waiver railroads did it:

The counter-intuitive figure is in FRA's next sentence: priority three locations "account for approximately 85 percent of the field-verified defects found as a result of continuous testing." They are the least likely to be real, yet they produce most of the confirmed defects because there are so many of them. That's why FRA kept the requirement to verify every suspect rather than letting owners drop low-confidence calls.

For a track owner, this shapes the verification workload. Most of a verification crew's day is spent on locations that turn out to be nothing, and the crew can't skip them. Lowering the number of benign suspects is the main lever on continuous-testing cost; our article on false calls in rail UT covers the causes, from weld flashing to bolt-hole geometry. Note that the 213.240(g) annual report counts defects and service failures but not benign suspects, so there is no federal false-call rate to compare against.

Key takeaway: plan verification capacity around the number of suspects, not the number of defects.

Does continuous rail testing find more defects?

Not on a single pass, but it lets owners test more often. The probes and physics are the same; the gain FRA cites comes from frequency, track time and comparison with earlier runs.

FRA's case for the 2020 rule rested on waiver experience. In the preamble it reported that:

• Norfolk Southern, testing continuously on limited territory from 2015 and expanding it several times, had 34% fewer main-line service failures in 2018 than in 2014.
• CSX, piloting continuous testing under waiver since 2009, had zero broken-rail-caused main-track accidents in 2019.
• The Long Island Rail Road, testing continuously under waiver since 2018, tripled its testing frequency "with no additional train delays."
• FRA reportable broken-rail-caused accidents fell 39% nationwide from June 2019 to May 2020.

Read these with care. They are FRA's summary of waiver results, not controlled comparisons. Rail renewal, grinding, tonnage and weather all move service failure counts, and the nationwide figure covers all railroads, not just continuous testers.

The case against moving too fast is on the record too. During the rulemaking the NTSB commented that "[u]ntil data from continuous rail testing can be collected, analyzed, and verified as beneficial to safety, the FRA should require that traditional stop-and-verify rail inspections" continue. Shipper associations worried that longer verification windows could leave "fatal flaws" unaddressed. FRA answered that the procedures and the most-restrictive-action backstop in (e)(5) were enough.

The detection limit itself doesn't change with the workflow. In a 2016 Technology Digest, TTCI wrote that conventional ultrasonic inspection "has proven effective at finding most rail flaws (about 90 percent)", leaving about 10% undetected during periodic inspection (TD-16-030). That figure applies to both methods. What continuous testing adds is what FRA calls "change detection": analysts compare the current run with past data from the same location, without a crew waiting on the track for the answer.

Key takeaway: more frequent tests and change detection are the safety argument for continuous testing; single-pass detection is unchanged.

What equipment do the two methods use?

Mostly the same: multi-channel ultrasonic roller search units, sometimes with induction. The 2020 preamble names ultrasonic as "the primary technology used, with induction technology currently used as a complementary system", at frequencies normally between 2.25 and 5.0 MHz. TTCI describes conventional detector cars as using "piezoelectric transducers housed in roller search units that run atop the railhead", with fixed-angle probes.

A typical probe set, which varies by system and should be confirmed for the equipment on your contract, includes:

Our guide to how ultrasonic testing works for railways explains each beam path.

What differs between the methods is the data path. A stop-and-verify car needs an operator qualified to interpret responses in real time. A continuous system needs reliable data transmission, a central analysis team qualified under 213.240(d)(2), and a geolocation system good enough to meet the "repeatable accuracy" standard in 213.240(f). Field verification in both cases is a hand test by a 213.238-qualified person, normally with a portable flaw detector and hand probes.

Phased array is emerging, not standard. TTCI built a phased array (PAUT) rail inspection prototype and reported in 2016 that it detected transverse defects under shells, weld defects and web defects at speeds up to 20 mph in tests at the Transportation Technology Center (TD-16-050). TTCI also noted that, like conventional probes, it can't inspect the rail base flanges, and that comparison with conventional testing in service was still future work. Treat any claim that production cars run "dynamic focal laws" as a vendor claim to check, not an industry norm.

Which method fits your track?

Continuous testing pays off on long, busy, well-documented segments; stop-and-verify suits short, quiet or defect-heavy track. No rule prefers one method, and many owners use both. This table sets out the factors:

Two edge cases deserve mention. A chase car test, one of FRA's four processes, sits between the two: a separate car verifies behind the test vehicle and keeps the lead car moving, without the 213.240 regime. And non-regulatory continuous runs under 213.240(e)(6) let an owner test extra mileage, such as yard track or mid-cycle checks, without the 36 and 84-hour windows, provided the test is designated and recorded as non-regulatory before it starts.

For tonnage-heavy territory, see our article on heavy-haul rail inspection; for the interval rules themselves, see how often rail must be tested.

Worked example: a 150-mile Class 4 line

On this example line, continuous testing cuts test-vehicle track time from 40 to 10 hours a year but needs two verification crews per test. The inputs below are planning assumptions for illustration, not industry norms. Replace them with your own data.

Inputs

Step 1: how often must the line be tested? Under 213.237(c), Class 4 track must be tested within 370 days or 30 MGT, whichever is shorter. At 45 MGT a year, 30 MGT arrives after 30 ÷ 45 × 365 = 243 days. Tonnage controls, so the owner plans two tests a year, about six months apart (about 22.5 MGT per interval), which leaves room for a retest after any invalid search.

Step 2: stop-and-verify track time.

• Moving time: 150 miles ÷ 15 mph = 10.0 hours
• Stop time: 60 stops × 10 minutes = 600 minutes = 10.0 hours
• Total occupancy: 20.0 hours per test, or 20 ÷ 5 = 4 work windows
• Average speed: 150 ÷ 20 = 7.5 mph

That 7.5 mph average sits inside the 6–8 mph that a 1999 university study reported as typical for stop-and-verify cars (MPC 99-106), so the inputs are realistic.

Step 3: continuous track time.

• Test-vehicle time: 150 ÷ 30 mph = 5.0 hours, or 1 work window
• Hand verification still has to happen: 60 suspects × 10 minutes = 10.0 hours of hand testing, now done by crews in short occupancies instead of inside the test-vehicle block

Step 4: can the crews meet the windows?

• All 60 suspects within 84 hours: 84 − 12 hours of analysis = 72 hours, which gives three day shifts
• One crew: 3 × 12 = 36 suspects, short of 60. Two crews: 3 × 2 × 12 = 72 suspects, enough
• The 4 priority one suspects within 36 hours: 36 − 12 = 24 hours, so they must be first on the first shift after analysis

Step 5: the annual picture.

An owner who keeps 40 hours of test-vehicle time could instead run four continuous tests a year (4 × 5 = 20 hours), shortening the tonnage between tests to about 11 MGT. That roughly matches the LIRR waiver result FRA reported, more frequent testing without added delay. The cost is 240 suspects a year to verify instead of 120.

Where the case flips. If the same line had worn, shelled rail producing 300 suspects per run (2 per mile), one test would need 300 ÷ 12 = 25 crew-shifts within three days, or about 9 crews (25 ÷ 3 = 8.3, rounded up). Stop-and-verify would also be slow: 300 × 10 minutes = 50 hours of stops plus 10 hours moving, or 60 hours (12 windows). On a line like that, rail renewal or grinding comes before any choice of test method, and a chase car may be the practical middle ground.

What drives the cost of each method?

Stop-and-verify costs track time; continuous testing costs people and systems behind the vehicle. FRA's regulatory evaluation of the 2020 rule estimated, over 10 years in 2019 dollars, costs of $27.4 million and cost savings of $149.3 million at a 7% discount rate, for net savings of $121.86 million. The largest savings line was "slow orders" at $141.3 million; continuous testing labor savings were $7.5 million. Those are FRA's industry-wide projections for owners who choose continuous testing, not a per-mile price for any one railroad.

For a single owner, the cost drivers are:

• Stop-and-verify: test-vehicle hours, the work windows and train delays they cause, and slow orders when a test can't finish before the 213.237 deadline
• Continuous: analysts and their 213.240(d)(2) qualification records, data transmission and storage, enough verification crews to clear the suspect list inside 84 hours, the 213.240(g) report, and the cost of most-restrictive-action restrictions on any suspect that misses its window
• Both: valid-search shortfalls under 213.237(g) and (h), which force a retest, a 25 mph limit or rail replacement regardless of method

Ask any testing contractor to price verification separately from test miles, and to state how the suspect list will be delivered, by when, and in what format.

What neither method can do

Neither method fixes the physical limits of ultrasonic testing. FRA's preamble notes that poor rail-head surface conditions such as shelling or corrugation, and conditions such as heavy lubrication or debris, can impair detection with any technology. Both methods use head-contact probes, so the base flanges outside the web aren't inspected; FRA describes coverage as the rail head and web "as well as the portion of the base directly beneath the web."

Other limits:

• Continuous testing doesn't change the test interval. FRA states that 213.240 does not modify "the required frequency of rail inspections" in 213.237 or 213.339.
• It doesn't change remedial action. Once a suspect is confirmed, the 213.113(c) table applies with the same timelines.
• It doesn't remove masking. A shelled head blocks the beam whether or not the car stops.
• A suspect is not yet a defect. Under (e)(4), a suspect isn't a 213.113(c) defect until a qualified person verifies it, but under (e)(5) a late suspect is treated as the worst case of its suspected type and size.
• The annual report doesn't measure accuracy. 213.240(g) counts tests, defects and service failures, not benign suspects or missed defects.
• Neither method replaces visual inspection under 213.233.
• Rail transit systems not connected to the general railroad system fall under FTA and state safety oversight, not Part 213, so 213.240 may not be the governing rule there.

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. On the short segments, yards and plant track those owners run, the practical question is usually how to fit a stop-and-verify test into the available work windows, and how to make sure the report lists every untested stretch and every defect's location, type and size. Whichever method you use, keep the test type, the verification times and the qualification records together, because those are what FRA will ask to see. To scope a test program, see our ultrasonic rail testing services or contact us.

Related reading: 49 CFR 213.237 rail flaw detection requirements · Ultrasonic railroad testing services explained · False calls in rail UT · Heavy-haul rail inspection · How ultrasonic testing works for railways

Frequently asked questions

Is continuous rail testing required by FRA?

Continuous testing is optional. Section 213.240(a) says a track owner "may elect" it to satisfy 213.237 or 213.339. FRA's 2020 rule states that owners who don't want to test continuously aren't required to.

How long does a track owner have to verify a suspect from a continuous test?

Within 84 hours of the indication, or within 36 hours if the suspected defect would need remedial action A, A2 or B. A broken rail with separation needs immediate protection. These windows replace the 4-hour and immediate limits in 213.113(b) only for a properly designated continuous test.

What happens if a suspect location isn't verified in time?

The owner must apply the most restrictive remedial action in the 213.113(c) table for the suspected type and size of defect, over enough track to cover the location, until it is field-verified (213.240(e)(5)).

Who can field-verify a suspect location?

A person qualified under 49 CFR 213.238, with documented training, at least 160 hours of supervised experience and written authorization from the employer for the equipment used. The analyst who reviewed the data needs a separate interpretation qualification under 213.240(d)(2).

Does continuous testing change how often rail must be tested?

The interval stays the same. Class 4 and 5 track still needs an internal rail inspection within 370 days or 30 MGT, whichever is shorter. Continuous testing makes it easier to test more often than the minimum, and extra non-regulatory runs can be designated under 213.240(e)(6).

Sources

• 49 CFR 213.240, Continuous rail testing, eCFR (text read on Cornell LII, October 2026)
• 49 CFR 213.113, Defective rails, eCFR
• 49 CFR 213.237, Inspection of rail, eCFR
• 49 CFR 213.238, Qualified operator, eCFR
• 49 CFR 213.233, Track inspections, eCFR
• Rail Integrity and Track Safety Standards, final rule, 85 FR 63362 (October 7, 2020), Federal Register via govinfo
• TTCI Technology Digest TD-16-030, High-Speed Rail Flaw Detection Using Phased Array Ultrasonics (Witte and Poudel, July 2016), MxV Rail
• TTCI Technology Digest TD-16-050, phased array rail inspection prototype (November 2016), MxV Rail
• MPC 99-106, rail flaw detection study (1999), Mountain-Plains Consortium / UGPTI