Eurostat Transport Database - Road and Rail Bulk Download

It is the road-and-rail half of the European Union's statistical record of movement, held as typed rows rather than browser tabs. The transport theme spans 667 tables, 182 of them rail, inside a database of roughly 1,400 - and the rail side alone divides into families worth knowing by name. rail_if_* covers infrastructure: track length split by electrification status, lines by track gauge, number of tracks, maximum permitted speed, ERTMS equipment and level crossings. rail_eq_* covers rolling stock by energy source and tractive power. rail_go_* and rail_pa_* carry goods and passengers; road_go_*, road_pa_* and road_eqs_* mirror the split on tarmac, down to vehicle fleets. tran_r_* then takes networks and vehicle stock down to NUTS 2 regions.

Scale next: individual tables hold anywhere from under 1,000 to about 670,000 values, and railway line-length series reach back to 1962 - few European series of any kind run that long on a harmonized basis. In our catalog it ranks third on the best heavy electrical equipment datasets list and tenth on the highways and railtracks one, pooled into both because network buildout is a demand signal for the equipment trades - see the heavy electrical equipment data hub and the highways and railtracks data hub.

Get a sample of this dataset cut to your families, countries and periods before anything else.

Two eras of the same table, captured during the August 2026 research pass:

DATAFLOW     : ESTAT:RAIL_IF_TRACKS(1.0)
freq         : A      unit : KM      tra_infr : RT_ELC
geo          : AL     TIME_PERIOD : 1990    OBS_VALUE : 0

DATAFLOW     : ESTAT:RAIL_IF_TRACKS(1.0)
freq         : A      unit : KM      tra_infr : RT_ELC
geo          : AT     TIME_PERIOD : 2019    OBS_VALUE : 7773

Read them together and the record's character shows. Albania's electrified-kilometre count sits at zero in 1990 and reaches a single kilometre by 1991 - a network at the very start of electrification. Austria reports 7,773 kilometres of energized track for 2019. Same ten columns, same key discipline, nearly thirty years apart: keys first, measure last, flags qualifying anything estimated or broken. That shape is why these rows stack into panels without reshaping, whether the panel spans six decades or six neighbours.

Ten documented fields define every observation, doing three jobs. Provenance: DATAFLOW names the agency, table and version a row belongs to, and LAST UPDATE stamps the release it came from - so nobody has to guess which vintage of the truth they are holding. Location: freq, unit, tra_infr, geo and TIME_PERIOD fix the cell - annual kilometres of electrified track in Germany in 2022, say. Resolution: OBS_VALUE carries the number, OBS_FLAG qualifies it (estimated values and breaks in series announce themselves rather than hiding), and CONF_STATUS marks the rare suppressed value.

tra_infr is the field the equipment trades care about: splitting TOTAL against RT_ELC and RT_NELC turns one infrastructure table into an electrification ledger. geo separates the union aggregate from national detail - continental benchmark versus one country's trajectory, same column. Definitions were verified against live responses during the August 2026 pass, and additional fields on request extend the frame with codebook lookups, regional cuts and modal joins rather than padding it with guessed columns.

• Geography - EU member states plus EFTA, EU candidate and other partner countries, with EU and euro-area aggregates riding beside the national rows; selected tables descend to NUTS 2 and NUTS 3 regions.
• Temporal - railway line-length series reach back to 1962, among the longest harmonized runs in European statistics; most rail and road series start between 1990 and 2005 and run to the latest reported year, 2024-2025 in the current release.
• Granularity - country x annual as the spine, with a few monthly and quarterly short-term series and regional breakdowns on the tran_r_* tables.

The practical consequence: a sixty-year infrastructure panel and a current-year read come off the same dimensional frame, so the long view and the short view reconcile without a mapping project between them.

API, files, or your warehouse. Daily, weekly, or hourly.

You pick the channel; the cadence follows the decision you are feeding. Load the historical depth once as a backfill - the 1962-onward railway runs reward it - then let new periods rotate into place on the schedule your systems expect. Deliveries keep the field dictionary above unchanged, ship the sample rows for validation, and resolve dimension codes against their codebooks, so joining onto your own country or asset master is a join statement rather than a decoding project. A sample cut to your named scope arrives first either way.

Six jobs the record settles outright:

1. Electrification demand signals - electrified versus non-electrified track kilometres by country turn railway investment into an addressable-market read for traction power and grid equipment.
2. Cross-country benchmarking - harmonized country-year grain makes member-state comparisons defensible without methodological surgery.
3. Infrastructure factor research - line-length series back to 1962 give long-horizon models a real deep panel; see the quant backtesting use case.
4. Citation-grade sourcing - official statistics attributable to the EU's statistical office stand in articles, filings and tenders; see citation-grade research.
5. Fleet and modal-shift tracking - vehicle fleets beside rail and road freight volumes make modal-share claims checkable; the tonne-kilometre glossary note unpacks the freight unit.
6. Regional siting analysis - NUTS 2 network and vehicle-stock tables locate demand where site decisions actually happen.

Market researchers and consultants benchmark member-state networks and fleet sizes with comparable official series; see market researchers working in heavy electrical equipment. Data scientists and ML engineers get long annual panels keyed by country and period that join outward onto production and installed-base features; see data scientists. Journalists, academics and students cite series reaching to 1962 with methodology attached; see journalists and academics. Investors and quant researchers backtest infrastructure and freight factors on six decades of railway runs; see investors and quants. Developers round it out: the ten-field frame drops into products without adaptation.

Notes worth holding beside it:

• Highway Performance Monitoring System (HPMS) - the American counterpart at route-segment grain; the trade-offs run side by side on the comparison page.
• FHWA Highway Statistics Series (1945-present) - US mileage, travel and finance tables with eight-decade depth.
• Geofabrik OpenStreetMap Data Extracts - feature-level geometry of the networks these tables measure.
• ORR Data Portal - UK Rail Statistics - regulator-grade depth when the question narrows to Britain.
• World Bank Railways Goods Transported Indicator - the global frame once the question outgrows the neighborhood.
• Eurostat Transport Theme - All Air Transport Datasets - the aviation wing of the same theme, delivered under its own record.

Upstream keeps only the latest edition of each table, which is why snapshot archiving rides in as an option rather than a nicety. Before requesting a sample, decide the families and the geography - rail infrastructure alone reads differently from the whole modal picture - and the Eurostat source profile covers the publisher behind all of it.