Is our certification program in order for the next audit?
Seven questions for anyone who certifies, qualifies or supervises the men who handle trains.
The question this Read asksWhat does a program nobody has read cost us a year, in our own handling events: damage, delay hours, claims and recrews?
Exposure Read 12 · The Qualified Man
Every railroad can produce a certification file for every man who handles its trains. Very few can produce a file that says what he was seen to do, on what territory, with what tonnage behind him, and who was watching.
That is not negligence. It is what happens when a program is built to satisfy an inspection rather than to produce an engineer. The classroom is documented because the classroom is easy to document. The hours that actually make a man — the trips where he got it wrong somewhere it did not matter, with somebody beside him who had got it wrong there first — leave almost no paper at all.
So the risk sits in the gap. A man is certified, the file is in order, the auditor is satisfied, and nobody in the building can tell you whether he has ever taken twelve thousand tons down that grade in the rain. The first time anyone asks that question out loud is usually after something has happened, when it is being asked by somebody who is not on your side.
The railroad certifies its engineers under 49 CFR Part 240 and its conductors under Part 242, each through a program FRA approves (240.103, 242.103), and trains its safety-related people under Part 243. FRA amended Part 243 by a final rule published 15 MAY 2026 and effective 14 JUL 2026 (91 FR 27849). The supervisor who gives the skill test and the annual observation need not be qualified on the territory; the test may be run on a simulator and the observation may be an event recorder download (49 CFR 240.127, 240.129). The absence that forces requalification is a number your own program sets (240.123(e); for conductors, 242.119(j)), and the skill test prescribes no count of supervised trips before it. What makes a man ready is left to the railroad, which is where these seven go.
Each rule cited here was read in its current published text on 13 NOV 2026, at the eCFR (current to 10 NOV 2026). The law changes after that date; read the current text before relying on any of it.
Run them yourself.
They are answerable from your own certification records, your own event recorder data and your own territory profiles. Run them yourself. They take an afternoon and they require nobody’s help, which is why they are yours to run.
Does the certification file show the man was evaluated, or that he attended?
Pull five files at random and read the entries. Count how many record an observed action, a condition, a correction made and the result. A file full of attendance and empty of observation certifies a schedule, not a man.
Who signs a man qualified, and what did that person actually watch?
Name the supervisor on the last ten qualifications. Then establish what each check ride covered — territory, tonnage, weather, time of day. A check ride run light power on a clear afternoon over the easiest subdivision qualifies a man to run light power on a clear afternoon over the easiest subdivision.
What territory is each man qualified on, and when did he last handle it?
Cross the qualification list against the actual assignment history. Territorial knowledge decays, and it decays fastest on the roads a man qualified on and then never ran. A man carrying a qualification he has not exercised in a year holds a piece of paper, not a memory of where the grade changes.
What does your simulator model, and what does it leave out?
Ask the vendor or your own program: is brake pipe propagation modeled to the rear of the consist, is slack action modeled, is the territory profile the real one? A simulator that models the sequence trains the sequence. What it cannot easily put into a man is the delay — the seconds between the handle and the response at the rear of a long train.
How many supervised trips before the first solo, and who chose the number?
Find the number and find its basis. If it is the same number for every man, every territory and every service, it was chosen for scheduling and not for competence. What matters is whether anybody is authorized to say a particular man is not ready yet, and whether that has ever happened.
When a handling event occurs, does the review reach the training that produced it?
Read the last three train-handling investigations. Count how many end at the man. An investigation that stops at the individual has found the last link and called it the cause. The useful question is whether the same handling would have come out of any other man the same program produced.
Who trains the trainers, and when were they last on the road themselves?
Establish the date each instructor last handled a train in revenue service on the territory he teaches. An instructor whose own road time is five years behind him is teaching the railroad as it was, and the men he produces will be corrected by the territory rather than by him.
A question you cannot answer from your own files is already an answer. Either the record does not exist, or it exists and nobody has read it since it was signed.
What a program nobody has read costs annually.
Set it beside the cost of reading it.
What arrives.
A program read against the territory
The certification, qualification and training files read against the roads the men actually run. What the file says each man can do, set beside what anyone has seen him do.
A handling event read in order
Train first and the file after the evidence, ending in a finding that names man, machine, territory or instruction, and what would have to change for it not to recur.
The written definition of a qualified man
Where none exists, a draft of one, built from the men whose judgment is already the standard, before they retire.
An answer on whether the program holds
Including, on some railroads, that it does. This read finds a program sound whenever it is.
And what it is not
Not a certification: the railroad certifies, and nothing here certifies or qualifies anyone. Not an opinion on any individual’s conduct: where a specific incident and person are the question, that is a different instrument, and the firm states which party has engaged it before the work begins.
The seven questions above are the test of the firm that wrote them. Run them, and judge the firm by what they find.
These seven are yours to run, and if the program holds you owe nobody anything.
If you need a second reader, one who does not report to the operating department, write and say what happened and when you need it.
Our bench of career railroaders brings better than 160 years of combined field experience. Nobody we send learns railroading on your property. That is a standard, not a résumé.
Write and say what you found.
Write and say what you found, or call 928.605.7984 and say it. Not an impression — what the seven returned on your own roster: your score, and which one your files could not answer. The one your files could not answer is the finding, and on most railroads it is the whole matter.
You will have a reply within one business day, from the principal and not from a form. Writing puts you on no list. If what comes back says you have no problem, you will be told so. What comes back sets the next stage, or the work stops there. The full Exposure Read, with its worksheet, comes from the same address.
This page is general information about railroad practice and the rules it cites, as they read on the date shown. It is not legal, engineering or financial advice, it does not replace counsel or an inspection of your property, and reading it does not engage SteelWheels.Co.
solutions@steelwheels.co · 928.605.7984 · What happens when you write back