# Weekly train speeds: coverage, sources and interpretation Reviewed September 20, 2026. The speed views cover eight Saturday–Friday periods, July 18–September 11, 2026. Charts use weeks ending July 24–September 11; official report dates are July 29–September 16. These are distinct dates. Sources were retrieved September 20, 2026. ## Availability Individual views include CSX, BNSF, Union Pacific, Norfolk Southern, CPKC and CN. The default CSX explorer separately retains four service measures and history since March 2017, with its [historical review and gaps](https://www.railstats.org/data/weekly-service-methodology.md). The comparison view supports one to four of CSX, BNSF, UP, NS and CPKC. It is a qualified descriptive overlay of reported line-haul speeds, not equivalent operating conditions, a service-quality ranking or end-to-end delivery speed. CN is available only individually. Earlier cross-railroad history and other measures are not cleared for these views. ## Reported fields and reconciliation Sources are the [STB Rail Service Data page](https://www.stb.gov/reports-data/rail-service-data/), its individual carrier workbooks and consolidated file through September 16, 2026. Each downloadable observation retains its original source URL, retrieval time, SHA-256 hash, worksheet and cell location, original value where stored as text, and consolidated reference value. CSX/BNSF/UP/NS/CPKC: Item 1, B13 in `Rail Service (Item Nos. 1-6)`, System, mph. CN: Item 1, B14 in `Service Metrics (items 1-6)`, System (U.S. Average), mph. NS contains numeric-text cells, explicitly parsed while retaining their original representation. BNSF's carrier-name header is blank; attribution follows the official STB BNSF archive URLs and matching compilation row. All 40 comparison observations match the compilation. Seven of eight CN observations match exactly. For the week ending August 14, the CN workbook reports 22.49 mph and the compilation reports 22.5 mph; these agree at one decimal. The individual view retains 22.49 and displays a range-specific precision note. No correction is fabricated. Train-type speeds are not averaged to produce system values. Underlying miles and hours are not supplied, so the published averages cannot be independently reconstructed from these weekly files. ## Railroad definitions and qualifications - [CSX](https://www.stb.gov/wp-content/uploads/CSX_Methodology.pdf): line-haul train speed excluding terminal time, local, yard, passenger and maintenance-of-way operations. - [BNSF](https://www.stb.gov/wp-content/uploads/BNSF_Methodology.pdf): train miles divided by operating hours, excluding terminal time and specified non-line-haul operations. - [UP](https://www.stb.gov/wp-content/uploads/UP-Methodology-EP-724-03.01.22.pdf): excludes intermediate terminal time and foreign trains, among other exclusions. Arrival/departure definitions changed in December 2019; this reviewed window is later. - [NS STB methodology](https://www.stb.gov/wp-content/uploads/NS_Methodology.pdf) and [current carrier definition](https://norfolksouthern.investorroom.com/weekly-performance-reports?cat=6): support terminal-time exclusion. NS filters calculated segment speeds above 80 mph and describes its operating population separately. - [CPKC](https://www.stb.gov/wp-content/uploads/CPKC-Methodology.pdf): excludes terminal dwell/station time and defines its system through train symbols. This is the STB-reported system, not a geography-matched network. Earlier CP and KCS figures are not spliced into this series. - [CN](https://www.stb.gov/wp-content/uploads/CN_Methodology.pdf): excludes time at crew-change locations, but does not clearly establish treatment of other terminal time for this U.S. series. CN is therefore shown individually with that uncertainty disclosed, and excluded from cross-road overlays. Its investor through-network speed is a separate measure and is not substituted. Complete historical continuity and a common all-six comparison start date remain unestablished. Differences in train population, traffic mix, route geography and terminal boundaries remain relevant even for the five eligible roads. The review supports descriptive use within this window; it does not certify all historical methods or establish causes for movements in speed. ## Calculations and controls The bounded speed views use original mph. Change equals the selected value minus the preceding displayed week's value. Calculations retain source precision; displayed speeds and changes round to two decimals. The first displayed week has no change. The separate historical CSX explorer can calculate a prior-week change outside the displayed window, as explained in its own methodology. Only the eight reviewed week endings are selectable in bounded views. Reversed ranges and unsupported comparisons are blocked, and unsupported requests clear previous results. One-week views show a point, not an inferred trend. CN data is supplied in the individual download only, not the comparison download. Exact values begin collapsed and retain source links. Download [individual speed observations](https://www.railstats.org/data/weekly-service-speed-individual.json) or [comparison observations](https://www.railstats.org/data/weekly-service-speed-comparison.json). These are derived website files; the source records link to the preserved official inputs. No smoothing, adjustment factor, interpolation, national average or composite score is applied. Individual explorers now also offer terminal dwell, cars online and trains held for these eight reviewed weeks. Their [separate methodology and source notes](https://www.railstats.org/data/weekly-service-individual-methodology.md) explain the additional measures. The comparison view remains train-speed only.