How train counts are reported on a crossing record, and why they can be old

By Tan Solutions · Published · Updated · 8 min readHow notes are written and corrected

The train boxes on the federal crossing form, the year each count carries, and why a recent revision date does not mean a recent count.

How many trains pass a crossing is one of the first questions people ask about it. Residents, planners and reporters all ask it. The federal crossing record has an answer, but the answer comes with a year, and that year is often older than the rest of the record. This note explains where the train figures sit on the record, who reports them, which deadline they follow, and how old they are across the country.

All figures in this note come from one reading of the federal inventory: the FRA dataset, data updated 12 November 2026, queried 13 November 2026. They are drawn differently from the site-wide count of 203,136 open crossings at track level in the inventory of 3 November 2026, so the two totals do not match exactly. Unless a sentence says otherwise, the figures cover the 203,162 records for crossings that are open and at track level, out of 438,949 rows in the dataset.

Where the train figures sit on the form

Every record is filed on the U.S. DOT Crossing Inventory Form, and the train figures sit in Part II, the railroad side of the form. The column descriptions of the FRA dataset on data.transportation.gov name each box. The ones that describe trains are these:

  • Box II.1.A, day through trains. The typical number of through trains between 6 a.m. and 6 p.m.
  • Box II.1.B, night through trains. The same count between 6 p.m. and 6 a.m.
  • Box II.1.C, switching trains. Trains that move cars around a yard or an industry rather than passing through.
  • Box II.1.D, transit trains. Trains of a transit service that use the crossing.
  • Box II.1.E, less than one movement per day. A box to check when trains are too few to count by the day, with a field for the number of trains per week.
  • Box II.2, Year of Train Count Data. The year the train figures describe.

Part I of the form adds one more figure, Box I.22, the number of passenger trains per day.

The counts are typical figures, not a log. A record that lists six day trains does not promise six trains on any given day. It describes the railroad's view of normal use in the year given in Box II.2. The second note in this series, reading a crossing record, field by field, walks through the rest of the form in order.

Who reports the counts

The duty to report sits with the railroad. The definitions in 49 CFR 234.401 name a primary operating railroad for each crossing, and that railroad files the record for it, train counts included.

Section 234.403 sets out how the data is sent. Each Class I railroad, the group of the country's large freight carriers, must submit its data electronically. A state agency may report on behalf of a railroad, and a parent corporation may report on behalf of its subsidiaries. So the name in the reporting agency field of a record can be a railroad, its parent company or a state, and any of them may be the source of the train counts you see.

Which deadline the counts follow

Two rules keep the inventory current, and the first note in this series, how the U.S. railroad crossing inventory is kept, explains both. For train counts, the point is which of the two applies.

The three-year update. Under section 234.409, the railroad must submit up-to-date and accurate data for each crossing it operates through at least every three years. Train counts are part of that data.

The three-month list. Section 234.411 names four changes that must reach the inventory within three months: the sale of all or part of a crossing, its closure, and, for public highway crossings, a change in the crossing surface or in the warning device. A change in the number of trains is not on that list.

So when a line gets busier or quieter, the new figure can wait for the next periodic update. And as the next sections show, a periodic update does not always bring a new train count with it.

How old the train counts are

Box II.2 is filled in on 186,889 of the 203,162 records. The years break down like this (FRA dataset, data updated 12 November 2026, queried 13 November 2026):

Year of Train Count Data (Box II.2)Records
202633,671
202543,427
202414,811
202335,399
2015 to 202258,612
Before 2015969
Blank16,273

Counts from 2024, 2025 and 2026 add up to 91,909 records. Counts from 2023 or later add up to 127,308. That leaves 58,612 records whose train counts are from 2015 to 2022, and 969 from before 2015.

A blank year is not the same as an old year. It means the box was not filled in, so the record does not say when its counts were taken. Of the 16,273 records with a blank year, 8,163 were last revised in 2015 or later and 8,110 before 2015. So a blank can sit on a recently revised record.

A recent revision date is not a recent count

The header of every record carries a revision date, Box A, the date the record was last updated. It is natural to read that date as the age of everything in the record, but for train counts that reading is often wrong. A railroad can revise a record for another reason, such as a new crossing surface or a corrected location, and leave the train figures as they were.

Comparing the two years on each record shows how often that happens (FRA dataset, data updated 12 November 2026, queried 13 November 2026):

  • train count year earlier than the year of the revision date: 111,141 records;
  • the same year: 75,747;
  • a later year: 1;
  • no train count year to compare: 16,273.

So more than half of the open crossings at track level have a train count older than their last revision. The size of the gap varies:

Train count older than the revision year byRecords
1 year37,168
2 years17,187
3 years15,013
4 years6,219
5 years5,482
6 years or more30,072

A gap of a year or two is ordinary: a railroad may count trains in one year and file early in the next. A gap of six years or more, which 30,072 records show, means the record has been revised while its train figures stayed in place for a long time. Revision dates are recent on most records: about 171,000 were last revised in 2024, 2025 or 2026, against 91,909 train counts from those same years.

None of this makes an old count wrong. A branch line that carried two trains a day in 2016 may still carry two. It means the record does not say so for the present, and the honest way to quote it is with its year: "two through trains a day, as counted in 2016".

Do public and private crossings differ?

The pattern is similar on both sides of the public and private line, with public records a little more recent. The note on public and private railroad crossings explains how the type is decided. Split by type (FRA dataset, data updated 12 November 2026, queried 13 November 2026):

  • Public crossings, 126,100 records. Train counts from 2023 or later: 83,038. Before 2023: 35,405. Blank: 7,657.
  • Private crossings, 76,867 records. From 2023 or later: 44,270. Before 2023: 24,176. Blank: 8,421.
  • No type recorded, 195 records. All have a blank train count year.

About two in three public records have a train count from 2023 or later, against a little under three in five private records. Private records also leave the year blank somewhat more often, about one in nine against about one in sixteen for public records. The differences are modest, and the train counts follow the same three-year clock on both sides.

Fewer than one train a day, and records with no trains

Many crossings see too few trains to count by the day. Box II.1.E is for them. Across the open crossings at track level (FRA dataset, data updated 12 November 2026, queried 13 November 2026):

  • the box for less than one movement per day is checked on 39,371 records;
  • it is marked as one or more movements per day on 148,647;
  • it is blank on 15,144.

The weekly count is filled in on 42,753 records. For a crossing with a train or two a week, that weekly figure is the one to read, because the daily boxes will show zero.

Zeros are common for that reason. On 49,130 records, the day through, night through and switching counts are all zero. Of those, 39,206 have the box for less than one movement per day checked, so the zeros simply say "not daily". Another 9,200 leave the box blank. The remaining 724 are marked as one or more movements per day while every daily count is zero, which does not add up on its face. Records like those are the ones to ask the reporting agency about.

A zero is also not the same as a closed crossing. A closure is one of the four changes due within three months, so a crossing that has been closed should show as closed, not as an open crossing with no trains. The note on closed and grade-separated railroad crossings explains how closures reach the record and describes the records this note leaves out of its counts.

What you can do with a train count

A train count is useful as long as it is read with its year and its source. A few habits help, whether you live by the crossing, plan for it or write about it.

  1. Read Box II.2 first. Quote the count with its year. If the year is blank, say the record does not give one.
  2. Compare it with the revision date. If the record was revised in 2026 and the count is from 2018, the record is current in other ways but its train figures are not.
  3. Add the boxes together carefully. Day and night through trains add up to the through total. Switching and transit trains are counted in their own boxes, and the passenger figure in Box I.22 sits in another part of the form; add figures together only when you know they describe different trains.
  4. Check the weekly field on quiet lines. Zeros in the daily boxes with the less-than-daily box checked mean a few trains a week, not none.
  5. Watch for a new count. The Rail Crossing Atlas page of weekly changes to the crossing records shows which records moved each week; in the week of 26 October 2026, 1,779 records had train counts updated. Its guide on reading a crossing record's history shows how a single record's figures have changed over the years.
  6. Ask the right party. The reporting agency named on the record filed the figures and can say whether a newer count exists.

Each record can be looked up by its crossing number in the FRA dataset, or on the Rail Crossing Atlas map; the Atlas page about the data describes where its figures come from. The counts on a record are the railroad's own account of normal use in a given year, and read with that year attached they are the best public answer there is.

Sources

Figures are as the source showed them on the date given. Rail Crossing Atlas figures come from the FRA crossing inventory of the date stated in the text.

  1. Crossing Inventory Data, Current: the FRA dataset and its column descriptions on data.transportation.gov. Read 13 November 2026.
  2. 49 CFR 234.401: definitions for crossing inventory reporting (Cornell Legal Information Institute). Read 13 November 2026.
  3. 49 CFR 234.403: submission of data to the crossing inventory (Cornell Legal Information Institute). Read 13 November 2026.
  4. 49 CFR 234.409: periodic updates to the crossing inventory (Cornell Legal Information Institute). Read 13 November 2026.
  5. 49 CFR 234.411: changes that must be reported within three months (Cornell Legal Information Institute). Read 13 November 2026.
  6. Rail Crossing Atlas: weekly changes to the crossing records. Read 13 November 2026.
  7. Rail Crossing Atlas guide: reading a crossing record's history. Read 13 November 2026.
  8. Rail Crossing Atlas: about the data. Read 13 November 2026.

More notes

All 13 notes

Look this up in Rail Crossing Atlas

Every U.S. crossing record, state by state, from the FRA inventory. Public records need no account.

Open Rail Crossing Atlas