Reference metadata describe statistical concepts and methodologies used for the collection and generation of data. They provide information on data quality and, since they are strongly content-oriented, assist users in interpreting the data. Reference metadata, unlike structural metadata, can be decoupled from the data.
Directorate for statistical data and studies (SDES)
1.2. Contact organisation unit
Departement for Transport Statistics
1.3. Contact name
Confidential because of GDPR
1.4. Contact person function
Confidential because of GDPR
1.5. Contact mail address
Services des Données et des Etudes Statistiques
Tour Sequoia
1 place Carpeaux
92055 Paris - La Défense Cedex
1.6. Contact email address
Confidential because of GDPR
1.7. Contact phone number
Confidential because of GDPR
1.8. Contact fax number
Confidential because of GDPR
2.1. Metadata last certified
3 December 2025
2.2. Metadata last posted
3 December 2025
2.3. Metadata last update
3 December 2025
3.1. Data description
Mandatory data collection at EU level is based on legal act, the Regulation EU 2018/643 and covers goods and passengers. Data is collected as following:
Annex I (annual data) –goods transport ,
Annex II (annual data)–passenger transport,
Annex III – (quarterly data) goods and passengers,
Annex IV and V( data every five years) regional statistics on goods and passengers and rail network, and
Annex VIII (annual data) goods and passengers transport for small undertakings.
The codes of regions used in the region-to-region statistics are indicated in Regulation (EC) 1059/2003 of the European Parliament and of the Council.
3.3. Coverage - sector
Railway undertakings providing transport of passengers or goods.
In 2024, 24 driver companies (included 7 foreign companies):
16 companies are only conducting freight transports;
6 companies are only conducting passengers transports;
2 company are conducting both types of transports.
Note that two passenger-transport companies merged in october 2023, and that a foreign freight company was unable to operate due to a line closure for the entire year.
We also collect data from 1 authorized candidate for freight transport.
Eurotunnel shuttle data (passengers) are also collected once a year.
We don't have direct access to information on rail waybill.
3.4. Statistical concepts and definitions
The main concepts used in rail domain are:
Rail passenger means any person, excluding members of the train crew, who makes a trip by rail. For accident statistics, passengers trying to embark/disembark onto/from a moving train are included.
Passenger-km means the unit of measure representing the transport of one passenger by rail over a distance of one kilometer. Only the distance on the national territory of the reporting country shall be taken into account.
Weight means the quantity of goods in tonnes (1 000 kilograms). The weight to be taken into consideration includes, in addition to the weight of the goods transported, the weight of packaging and the tare weight of containers, swap bodies, pallets as well as road vehicles transported by rail in the course of combined transport operations. If the goods are transported using the services of more than one railway undertaking, when possible the weight of goods shall not be counted more than once.
Tonne-km means the unit of measure of goods transport which represents the transport of one tonne (1 000 kilograms) of goods by rail over a distance of one kilometer. Only the distance on the national territory of the reporting country shall be taken into account.
Train means one or more railway vehicles hauled by one or more locomotives or railcars, or one railcar travelling alone, running under a given number or specific designation from an initial fixed point to a terminal fixed point. A light engine, that is to say, a locomotive travelling on its own, is not considered to be a train.
Train-km means the unit of measure representing the movement of a train over one kilometre. The distance used is the distance actually run, if available, otherwise the standard network distance between the origin and destination shall be used. Only the distance on the national territory of the reporting country shall be taken into account.
3.5. Statistical unit
The data used in the domain are collected at railway companies level.
3.6. Statistical population
Data on passenger and freight transport are collected from the railway undertakings operating at national territory within thresholds mentioned in the rail regulation. Data on traffic are collected from infrastructure managers and railway undertakings.
Passengers:
Numbers of passengers in high speed train travels and most long distance travels are derived from ticket information as reservation is compulsory even if the client have a seasonal ticket. The data may nevertheless be revised from one quarter to another (taking into account cancellations, etc.)
For commuting trips, especially within the Paris area, numbers of passengers are model- based estimates, and can be revised from one quarter to another.
The model used by the rail way company to estimates the numbers of passengers within the Paris area incorporates three types of information:
Actual counts of loading and unloading of passengers in rail stations. The counting operations are occurring at a given date, different for every station but each and every station is supposed to host a counting operation in a time span of five years or so.
Number of ticket validations, including seasonal tickets, in all the stations of the network, for each and every day
Passengers Mobility Survey.
3.7. Reference area
Operational railway network on national territory.
3.8. Coverage - Time
Data on passengers and goods are covered from 2004 onwards.
3.9. Base period
Not applicable.
The volume and performance of rail freight traffic are measured in tonnes (mass) and tonne-kilometres. Passenger transport by rail is measured in the number of passengers and in passenger-kilometres. Information on the number of train kilometres is also available. Traffic flows on the rail network are measured in number of trains - passenger, freight and others (optional).
The tables consist mostly of annual data. There are some tables providing quarterly and quinquennial (every five years) data.
According to the Rail regulation (EU) 2018/643 data is collected as following: Annual data
Annex I –goods transport collected for a reference period of one year with a deadline for data transmission as 5 months after the reference periodreference period,
Annex II –passenger transport, yearly data with eight months deadline after the end of reference period,
Annex III – quarterly data for goods and passengers with a deadline of three months after the end of the reference period,
Annex IV and V data every five years on regional statistics on goods and passengers and rail network with deadline of 12 respectively 18 months after the end of reference period and
Annex VIII goods and passengers transport for small undertakings with 5 respectively 8 months deadline after the end of reference period.
6.1. Institutional Mandate - legal acts and other agreements
Law no.51-711 of 7 June 1951 (amended) on legal obligation, coordination and confidentiality in the field of statistics defines statistical confidentiality, its limits and conditions of application.
Article L.311-8 of the French Code of Public Relations between thepublic and general government
European level:
(recital 24 and Article 20(4)) of 11 March 2009 (OJ L 87, p. 164), stipulates the need to establish common principles and guidelines ensuring the confidentiality of data used for the production of European statistics and the access to those confidential data with due account for technical developments and the requirements of users in a democratic society.
7.2. Confidentiality - data treatment
The flag of confidentiality indicates if data provided is considered confidential by the Member State or not (confidentiality laws no 322/97, Art. 13 §1 and no 1588/90, Art. 2).
Council Regulation (EC) no 322/97, Art. 13 § 1:
Data used by the national authorities and the Community authority for the production of Community statistics shall be considered confidential when they allow statistical units to be identified, either directly or indirectly, thereby disclosing individual information. To determine whether a statistical unit is identifiable, account shall be taken of all the means that might reasonably be used by a third party to identify the said statistical unit.
Council Regulation (EC) no 1588/90, Art. 2:
For the purposes of this Regulation, the terms given below are defined as follows:
confidential statistical data: data declared confidential by the Member States in line with national legislation or practices governing statistical confidentiality;
national authorities: national statistical institutes and other national bodies responsible for the collection and use of statistics for the Communities;
- information on the private lives of natural persons: information on the private and family lives of natural persons as defined by national legislation or practices in the various Member States;
use for statistical purposes: use exclusively for the compilation of statistical tables or for statistical economic analyses; may not be used for administrative, legal or tax purposes or for verification against the units surveyed;
statistical unit: basic unit covered by statistical data transmitted to the Statistical Office of European Community (Eurostat);
direct identification: identification of a statistical unit from its name or address, or from an officially allocated and published identification number;
indirect identification: possibility of deducing the identity of a statistical unit other than from the information mentioned in point 6;
officials of the SOEC: officials of the Communities, within the meaning of Article 1 of the Staff Regulations of Officials of the European Communities, working at the SOEC;
other staff of the SOEC: servants of the Communities, within the meaning of Articles 2 to 5 of the Conditions of Employment of Other Servants of the European Communities, working at the SOEC;
dissemination: supply of data in any form whatever: publications, access to databases, microfiches, telephone communications, etc.
The national authorities shall make a request to such undertakings for permission to disclose the required data and shall inform Eurostat of the result of this request when data are transmitted to Eurostat. Confidentiality of the data is based on Member State's judgment.
Council Regulation (EC) no 1588/90, Art. 3:
The national authorities shall be authorised to transmit confidential statistical data to the SOEC.
National rules on statistical confidentiality may not be invoked to prevent the transmission of confidential statistical data to the SOEC where an act of Community law governing a Community statistic provides for the transmission of such data.
Transmission to the SOEC of confidential statistical data on the structure and activity of undertakings, collected before the entry into force of this Regulation, must be carried out in accordance with the rules and practices on statistical confidentiality in force in the Member States.
Transmission to the SOEC of confidential statistical data, within the meaning of paragraph 2, shall be carried out in such a way that statistical units cannot be directly identified. This does not preclude the admissibility of more far-reaching transmission rules in accordance with the legislation of the Member States.
National authorities shall not be obliged to transmit information on the private lives of natural persons to the SOEC if the information transmitted could enable those persons to be identified either directly or indirectly.
8.1. Release calendar
According to he Directive 2009/42/EC of the European Parliament and the council of 6 May 2009, the transmission of the results shall take place within five months of the end of the period of observation for data of quarterly periodicity, within five months for data of annual goodsperiodicity, within eight months for data of annual passengers periodicity, and within two years for the quinquennial periodicity.
Eurostat shall disseminate appropriate statistical data with a periodicity comparable to that of the results transmitted.
Dissemination to EUROSTAT: the tables consist mostly of annual data. There are some tables providing quarterly and quinquennial (every five years) data.
According to the Rail regulation (EU) 2018/643 data is collected as following: Annual data
Annex I –goods transport collected for a reference period of one year with a deadline for data transmission as 5 months after the reference periodreference period,
Annex II –passenger transport, yearly data with eight months deadline after the end of reference period
Annex III – quarterly data for goods and passengers with a deadline of three months after the end of the reference period,
Annex IV and V data every five years on regional statistics on goods and passengers and rail network with deadline of 12 respectively 18 months after the end of reference period and
Annex VIII goods and passengers transport for small undertakings with 5 respectively 8 months deadline after the end of reference period.stat
Data are collected directly from Railway Companies in compliance with the methodology and classifications specified in the legal framework on European Level.
11.1. Quality assurance
Quality assurance framework for the RailwayTransport Statistics is based on the ES CoP, the European Statistics Code of Practice adopted by the Statistical Programme Committee on 24 February 2005 and revised by the European Statistical System Committee in September 2011 and in November 2017. The European Statistics Code of Practice sets the standard for developing, producing and disseminating European statistics, along the lines of the institutional environment, statistical processes and statistical output.
Statistical authorities, comprising Eurostat, the National Statistical Institutes (NSIs) and Other National Authorities (ONAs) responsible for the development, production and dissemination of European Statistics, are strongly committed to quality – this commitment and high quality awareness are clearly expressed in the Quality Declaration of the ESS that is also included in the Preamble of the CoP.
The Quality Assurance Framework of the European Statistical System (ESS QAF) complements and breaks further down the Code of Practice.
There is a set of validation rules and quality checks put in place, which detect various types of issues. In case of any issues detected, the data providers are contacted by Eurostat to provide explanations or/and revise the data accordingly.
The data providers then check the data, and return to the Railway companies if needed.
12.1. Relevance - User Needs
The Railway Transport Statistics covers the information needs of different groups of users, among wich are different organizations, Public Administration, territorial administrations, private companies, media, researchers and universities, private enterprises and individuals.
12.2. Relevance - User Satisfaction
Eurostat conducts two types of evaluations that assess Eurostat's performance in general, and one of them are the User Satisfaction Surveys. The report for 2024.
SDES is committed in understanding who the users of the statistics it produces are, what the data needs are, whether they match production and if the statistics produced fulfill the needs of users.
To this aim SDES is constantly in contact with key users of the statistics (among them national authorities, professional organization for French transport, public statistics, ...) discussing results and new needs from both sides.
The overall accuracy is considered to be good, notably when data are compared with other sources, such as statistics compiled by ART (Autorité de Régulation des Transports).
13.2. Sampling error
Not applicable for Rail transport collection.
13.3. Non-sampling error
Coverage errors
When a Railway Company doesn't respond in times, estimations are done. It's rare and usually concern low freight companies.
When a new Railway Company starts its activity on national territory, some difficulties to contact it may appears, and concepts may take time to be well understood.
Measurement errors
Some marginal errors may occur as a result of codification errors by Railway Companies. For example, a large share of so-called “other” instead of specific codes for type of goods. The distance reference can be different in the information system from a Railway Company to another. This is also the case for the UIC codes that we use to create the NUTS variables (some that don't exist, or are false), some of them are corrected when possible. The Railway Companies can have difficulties to collect from their clients all the information the statistics need (number of UTI, ...).
Processing errors
Some ajustements has to be done when a Railway Company can't give all the information the statistics need, or when there is an obvious error: in that case, confirmation is requested from the Railway Company, which can return corrected data, or adjustments are applied.
14.1. Timeliness
According to he Directive 2009/42/EC of the European Parliament and the council of 6 May 2009, the transmission of the results shall take place within five months of the end of the period of observation for data of quarterly periodicity, within five months for data of annual goodsperiodicity, within eight months for data of annual passengers periodicity, and within two years for the quinquennial periodicity.
Eurostat shall disseminate appropriate statistical data with a periodicity comparable to that of the results transmitted.
14.2. Punctuality
The deadliness for transmission are usually respected. Sometimes (rare), the companies can be late, so we estimate their part, then we revised the data once we receive them.
15.1. Comparability - geographical
Geographical comparability is obtained via the application of common concepts and validation rules. The harmonised interpretation of the methodology is enhanced by the organisation of regular meeting of the Working Group on Rail Transport Statistics, where all reporting countries are represented.
However, an effort remains to be made, with the collaboration of other countries, on the mirror checks provided by Eurostat.
15.2. Comparability - over time
There is one significant break in the series of Railway Freight statistics :
A significant operator is missing before 2015 (from 2005).
15.3. Coherence - cross domain
Cross domain data comparison (for different modes of transport) is not always possible. For instance, maritime transport statistics are not yet teritorialised like the other modes of transport (IWW, road, railway, aviation). A consultation of the member states has been conducted by Eurostat in 2020 within the framework of the WG on Maritime trasnport staistics in order to get their views on the use of a specific method based on a distance matrix and the Exclusive Economic Zones (EEZ).
Once the use of this method will be approved it will be possible to include maritime transport in modal split indicators and thus enable cross domain comparisons.
15.4. Coherence - internal
Each set of results is internally made consistent. For example, quarterly data are made consistent with annual data during annual collection.
The most time-consuming may be the various exchanges with the railway companies, when necessary.
17.1. Data revision - policy
SDES revision policy is reported at the following website.
17.2. Data revision - practice
The datasets reported quaterly can be revised once a year when annual datasets are transmitted.
The datasets reported annualy can be revised when necessary, or on the occasion of Annex IV (every five years).
18.1. Source data
Data are collected by SDES using railway companies information system as a data source.
In 2024, 24 driver companies (included 7 foreign companies):
16 companies are only conducting freight transports;
6 companies are only conducting passengers transports;
2 company are conducting both types of transports.
Note that two passenger-transport companies merged in october 2023, and that a foreign freight company was unable to operate due to a line closure for the entire year.
We also collect data from 1 authorized candidate for freight transport.
Eurotunnel shuttle data (passengers) are also collected once a year.
We don't have direct access to information on rail waybill.
18.2. Frequency of data collection
Quaterly, annual and quinquennial.
18.3. Data collection
Data collection using a template spreadsheet (one for the quaterly data, one for the annual data) and transmission using a dedicated functionnal mailbox with restricted access, some using crypted Excel files, some chose not to encrypt files.
18.4. Data validation
Data are validated using automated programs, which produce some outputs : modalities of variables, consistency over time, intra datasets, interdatasets. Some manual checks can also be done if necessary.
If surprising developments or declarations are detected, a return is made to the railway undertakings concerned for validation or correction.
The companies are reporting data using a classification scheme coherent with the requirement of the regulation, although it might be at a more detailed level, and some ajustements can be done (wrong code or wrong format, …), or some variables have to be created (NUTS0-2 or TRANSPORT_TYPE for example).
The type of goods classification required by the legislation is the NST 2007 at the four digit level, but most of respondents gives three or two digit level.
For origin and destination, depending on the transport (National or International), we can have either a UIC code, or Nuts3 or Nuts 2 code (but we mostly have UIC code since 2020 data).
For freight transport we trust the reporting distances to be actual distances as the companies are using a distance matrix. We have macro controls, and controls on changing company distances.
For passengers transport, distances are actual distances for travel on high speed trains (for a few lines it’s the commercial distance), and most of long distance travels. For commuting trips, especially within the Paris area, distances are model- based estimates.
18.5. Data compilation
When compiling the data of different railway companies, estimation might be used when necessary, for example when a companies don't respond in time. Generally, these estimations concern low weight railway undertakings.
Automatic adjustments, general or specific (terms, values, ...), are also applied, and some variables are created.
For the quarterly data, the type of transport is fixed in the file, and aggregated data are completed by the reporting companies.
For the annual data, the type of transport is calculated thanks to the actual origin and actual destination at the UIC (mostly), Nuts1, Nuts2 or Nuts3 level that is informed by the reporting companies. There are no big problems in assigning the type of transport. Certain border points are corrected when identifiable.
18.6. Adjustment
Automatic adjustments, general or specific (terms, values, ...), are also applied, and some variables are created.
Improvement on freight tables A4, A7 and A8.
Tables A7 and A8 are the most difficult to produce. Our data collection regulation does not provide the railways companies to report the number of ITUs. We have recently improve with the companies, but there’s still some work to do for a correct quality. With the 2023 collection, we have authorized publication on the Eurostat database. The quality is estimated to be correct from 2020 to 2024, but the 2016-2019 data should still be taken with caution.
For passengers, we used to use data from Luxembourg (from/to France) to fill a lack in the declaration of the main company, on a partnership with a Luxembourg railway company (2020-2022). Since 2023, this lack has been fixed.
Mandatory data collection at EU level is based on legal act, the Regulation EU 2018/643 and covers goods and passengers. Data is collected as following:
Annex I (annual data) –goods transport ,
Annex II (annual data)–passenger transport,
Annex III – (quarterly data) goods and passengers,
Annex IV and V( data every five years) regional statistics on goods and passengers and rail network, and
Annex VIII (annual data) goods and passengers transport for small undertakings.
3 December 2025
The main concepts used in rail domain are:
Rail passenger means any person, excluding members of the train crew, who makes a trip by rail. For accident statistics, passengers trying to embark/disembark onto/from a moving train are included.
Passenger-km means the unit of measure representing the transport of one passenger by rail over a distance of one kilometer. Only the distance on the national territory of the reporting country shall be taken into account.
Weight means the quantity of goods in tonnes (1 000 kilograms). The weight to be taken into consideration includes, in addition to the weight of the goods transported, the weight of packaging and the tare weight of containers, swap bodies, pallets as well as road vehicles transported by rail in the course of combined transport operations. If the goods are transported using the services of more than one railway undertaking, when possible the weight of goods shall not be counted more than once.
Tonne-km means the unit of measure of goods transport which represents the transport of one tonne (1 000 kilograms) of goods by rail over a distance of one kilometer. Only the distance on the national territory of the reporting country shall be taken into account.
Train means one or more railway vehicles hauled by one or more locomotives or railcars, or one railcar travelling alone, running under a given number or specific designation from an initial fixed point to a terminal fixed point. A light engine, that is to say, a locomotive travelling on its own, is not considered to be a train.
Train-km means the unit of measure representing the movement of a train over one kilometre. The distance used is the distance actually run, if available, otherwise the standard network distance between the origin and destination shall be used. Only the distance on the national territory of the reporting country shall be taken into account.
The data used in the domain are collected at railway companies level.
Data on passenger and freight transport are collected from the railway undertakings operating at national territory within thresholds mentioned in the rail regulation. Data on traffic are collected from infrastructure managers and railway undertakings.
Passengers:
Numbers of passengers in high speed train travels and most long distance travels are derived from ticket information as reservation is compulsory even if the client have a seasonal ticket. The data may nevertheless be revised from one quarter to another (taking into account cancellations, etc.)
For commuting trips, especially within the Paris area, numbers of passengers are model- based estimates, and can be revised from one quarter to another.
The model used by the rail way company to estimates the numbers of passengers within the Paris area incorporates three types of information:
Actual counts of loading and unloading of passengers in rail stations. The counting operations are occurring at a given date, different for every station but each and every station is supposed to host a counting operation in a time span of five years or so.
Number of ticket validations, including seasonal tickets, in all the stations of the network, for each and every day
Passengers Mobility Survey.
Operational railway network on national territory.
The tables consist mostly of annual data. There are some tables providing quarterly and quinquennial (every five years) data.
According to the Rail regulation (EU) 2018/643 data is collected as following: Annual data
Annex I –goods transport collected for a reference period of one year with a deadline for data transmission as 5 months after the reference periodreference period,
Annex II –passenger transport, yearly data with eight months deadline after the end of reference period,
Annex III – quarterly data for goods and passengers with a deadline of three months after the end of the reference period,
Annex IV and V data every five years on regional statistics on goods and passengers and rail network with deadline of 12 respectively 18 months after the end of reference period and
Annex VIII goods and passengers transport for small undertakings with 5 respectively 8 months deadline after the end of reference period.
The overall accuracy is considered to be good, notably when data are compared with other sources, such as statistics compiled by ART (Autorité de Régulation des Transports).
The volume and performance of rail freight traffic are measured in tonnes (mass) and tonne-kilometres. Passenger transport by rail is measured in the number of passengers and in passenger-kilometres. Information on the number of train kilometres is also available. Traffic flows on the rail network are measured in number of trains - passenger, freight and others (optional).
When compiling the data of different railway companies, estimation might be used when necessary, for example when a companies don't respond in time. Generally, these estimations concern low weight railway undertakings.
Automatic adjustments, general or specific (terms, values, ...), are also applied, and some variables are created.
For the quarterly data, the type of transport is fixed in the file, and aggregated data are completed by the reporting companies.
For the annual data, the type of transport is calculated thanks to the actual origin and actual destination at the UIC (mostly), Nuts1, Nuts2 or Nuts3 level that is informed by the reporting companies. There are no big problems in assigning the type of transport. Certain border points are corrected when identifiable.
Data are collected by SDES using railway companies information system as a data source.
In 2024, 24 driver companies (included 7 foreign companies):
16 companies are only conducting freight transports;
6 companies are only conducting passengers transports;
2 company are conducting both types of transports.
Note that two passenger-transport companies merged in october 2023, and that a foreign freight company was unable to operate due to a line closure for the entire year.
We also collect data from 1 authorized candidate for freight transport.
Eurotunnel shuttle data (passengers) are also collected once a year.
We don't have direct access to information on rail waybill.
Dissemination to EUROSTAT: the tables consist mostly of annual data. There are some tables providing quarterly and quinquennial (every five years) data.
According to the Rail regulation (EU) 2018/643 data is collected as following: Annual data
Annex I –goods transport collected for a reference period of one year with a deadline for data transmission as 5 months after the reference periodreference period,
Annex II –passenger transport, yearly data with eight months deadline after the end of reference period
Annex III – quarterly data for goods and passengers with a deadline of three months after the end of the reference period,
Annex IV and V data every five years on regional statistics on goods and passengers and rail network with deadline of 12 respectively 18 months after the end of reference period and
Annex VIII goods and passengers transport for small undertakings with 5 respectively 8 months deadline after the end of reference period.stat
According to he Directive 2009/42/EC of the European Parliament and the council of 6 May 2009, the transmission of the results shall take place within five months of the end of the period of observation for data of quarterly periodicity, within five months for data of annual goodsperiodicity, within eight months for data of annual passengers periodicity, and within two years for the quinquennial periodicity.
Eurostat shall disseminate appropriate statistical data with a periodicity comparable to that of the results transmitted.
Geographical comparability is obtained via the application of common concepts and validation rules. The harmonised interpretation of the methodology is enhanced by the organisation of regular meeting of the Working Group on Rail Transport Statistics, where all reporting countries are represented.
However, an effort remains to be made, with the collaboration of other countries, on the mirror checks provided by Eurostat.
There is one significant break in the series of Railway Freight statistics :
A significant operator is missing before 2015 (from 2005).
