Methodology

Sources

The figures on these pages come from the vehicle register of the RDW (Rijksdienst voor het Wegverkeer, the Dutch road transport agency), published as open data and read through RDW's Socrata API. Two datasets are used: Gekentekende voertuigen (m9d7-ebf2) for the vehicles themselves and Gekentekende voertuigen brandstof (8ys7-d773) for their fuel rows. Both are published under CC0 and are credited to "Team Open Data RDW". The register snapshot behind this export was taken on 2026-09-07; the latest complete month is 2026-08 and the site shows the latest 36 months.

The official series for new passenger cars is CBS StatLine table 85898NED (Statistics Netherlands, CC BY 4.0). CBS publishes no make and no model dimension, so it is used as a monthly check on the new-car counts rather than as a source. BOVAG and RAI Vereniging publish make-level registration figures, but under no reuse licence, so they are not used here.

What is counted

• Passenger cars: RDW's voertuigsoort = 'Personenauto', which in the register is exactly European vehicle category M1 (verified against europese_voertuigcategorie).
• The month is the month of the car's first Dutch registration (datum eerste tenaamstelling in Nederland), so a registration counts once, when the car first enters Dutch traffic under a keeper's name.
• Cars that were never put in a keeper's name — dealer stock, about 0.2 % of the register — carry no such date and are left out.

New cars versus used imports, and the CBS check

A car is counted as new when its first admission anywhere (datum eerste toelating) falls on the same day as its first Dutch registration; otherwise it is a used import. RDW records no "new" flag of its own, so this day-equality rule is the definition, and it is checked every month against CBS. Over full years the rule lands within a percent of the official series: 2024 came out 0.6 % below CBS and 2025 0.3 % above, and every month from January 2024 onward sits within 2.6 % of it. A freshly frozen month whose gap exceeds 1.5 % stops the pipeline — 3 % for a month filled in by the initial backfill or one CBS still calls provisional (which, since CBS has marked every month of the last three years provisional, is the band that applies in practice), and 5 % for a month frozen more than three years after it ended, where scrappage the register cannot show has already eroded the count — so a figure that drifts away from the official series is never published.

Months CBS has not published yet are not listed; CBS publishes about three months after the quarter.

Freeze and survivorship

The RDW register is a live file: it holds the cars registered today, not a record of what was registered in a past month. Each month is therefore measured once, on the first snapshot whose dataset truth date is at least three days past the month's last day, and is then frozen and never recomputed. The pairing of month and snapshot is recorded in the pipeline's frozen_months table, so every published figure can be traced to the snapshot it was measured on. The freeze decision uses the dataset's own truth date rather than the calendar, because RDW's vehicle datasets occasionally stop refreshing for several days.

Freezing early keeps the loss small but not zero: a car that has been exported or scrapped since its registration is gone from the register. Dutch exports stay in the file for two years, so within the site's window only scrapped cars actually vanish. Measured on 2026-09-07, the snapshot still held 100.6 % of CBS's 2025 new cars, 99.6 % of 2024, 98.6 % of 2023 and 97.6 % of 2022 — so across the 36 months shown the shortfall is at most about 1.4 %, small enough that year-on-year comparisons here are meaningful.

Fuel groups and hybrid classes

A car's powertrain is derived from its RDW fuel rows, one row per fuel the vehicle can use, each carrying the car's hybrid class (klasse_hybride_elektrisch_voertuig) repeated on it. The groups are, in the order they are tested:

• Other first for hydrogen and the fuel-cell classes OVC-FCHV and NOVC-FCHV: a fuel-cell car's first fuel row is Elektriciteit and its second Waterstof, so it has to be recognised before the battery-electric test or a Mirai would be counted as a BEV. Ethanol and the roughly 0.1 % of cars with no fuel row also land here.
• Battery electric: every one of the car's fuel rows is Elektriciteit.
• Plug-in hybrid: hybrid class OVC-HEV (outside-vehicle chargeable). Hybrid: class NOVC-HEV — a hybrid without a plug. This is the type-approval class, which does not separate a full hybrid from a mild (48 V) hybrid: a Suzuki Swift, a Ford Puma or a Volkswagen Golf eTSI carries NOVC-HEV just as a Toyota Yaris does, so the hybrid group includes mild hybrids. That matches the BOVAG/RAI fuel split (August 2026: 11,339 hybrids and only 1,613 petrol cars among 26,455 new cars) but makes the share higher than a full-hybrid-only count would be.
• Petrol, diesel and gas from the first fuel row: Benzine, Diesel, and LPG / CNG / LNG respectively. Any car with Benzine as its first row counts as petrol whether an LPG or CNG tank was fitted at the factory or later — Dacia's bi-fuel ECO-G models register that way — so the gas group holds almost no new cars (one in the 36 months to August 2026, where BOVAG/RAI count about 190 LPG cars a month) and is essentially a used-import group.

CO₂ and import age

CO₂ is the type-approval WLTP value emissie_co2_gecombineerd_wltp, falling back to co2_uitstoot_gecombineerd (NEDC) where no WLTP value is recorded. It is taken from whichever of the car's fuel rows carries a value rather than from the first one, because a full hybrid's figure sits on its petrol row while its first row is Elektriciteit. Battery-electric cars count as 0 g/km.

Plug-in hybrids have no CO₂ value in RDW's fuel dataset: emissie_co2_gecombineerd_wltp is empty on every fuel row of every plug-in hybrid. RDW publishes their figure in a different field — emis_co2_gewogen_gecombineerd_wltp, the weighted combined WLTP value — which this pipeline does not read yet. The published average therefore covers only the month's new cars that have a value: battery-electric cars at 0 g/km, petrol, diesel and hybrids without a plug, with plug-in hybrids left out entirely. Since the omitted cars would carry weighted figures well below a petrol car's, the average shown here runs higher than a new-fleet average that included them; read it as the average of the new cars that have a type-approval figure rather than as the average of every new car. Adding the weighted field is a planned change.

Age at import is the first Dutch registration minus the first admission anywhere, in years (days ÷ 365.25), grouped into brackets under 3, 3–5, 6–10 and over 10 years. RDW does not record the country a car was imported from, so there is no origin table on the imports page.

KPI definitions

Registrations

All passenger cars (voertuigsoort Personenauto, equal to category M1) first registered in the Netherlands in the month, from the RDW vehicle register: new cars plus used imports. Cars that were never put in a keeper's name are excluded.

New cars / New share

Cars whose first admission anywhere (datum eerste toelating) is the same day as their first Dutch registration. New share = new cars ÷ all registrations. Checked every month against CBS StatLine 85898NED.

Used imports / Avg import age (yrs)

Cars first admitted abroad before their Dutch registration. Average import age = the time between the two dates, in years.

BEV share

Battery-electric cars as a share of all registrations in the month, new and used together. A car is battery electric when every one of its RDW fuel rows is Elektriciteit. PHEV = the OVC-HEV hybrid class, counted separately.

New-car BEV share

Battery-electric cars as a share of new registrations only.

Hybrid share

Hybrids without a plug: RDW's NOVC-HEV hybrid class, as a share of all registrations in the month. The class is the type-approval one and covers full hybrids and mild (48 V) hybrids alike, so a Suzuki Swift or a Volkswagen eTSI counts here, as it does in the BOVAG/RAI fuel split. Plug-in hybrids (OVC-HEV) are counted separately.

Avg CO₂, new (g/km)

Mean type-approval CO₂ of the month's new cars: the WLTP value where the register records one, else the NEDC value, taken from whichever fuel row carries it. Battery-electric cars count as 0 g/km. Plug-in hybrids have no CO₂ value in RDW's fuel dataset (RDW publishes theirs in a separate weighted-WLTP field this pipeline does not read yet), so they are left out of the average entirely and the figure shown runs higher than a new-fleet average that included them.

MoM / YoY / YTD

Change versus the previous month, versus the same month a year earlier, and the calendar year to date with its change against the same months a year earlier. Percentages are shown only when the comparison month has at least 30 registrations, or 10 on the electric tables.

Normalisation

The register's merk and handelsbenaming are spelled inconsistently and, for some makes, at trim level rather than model level, so both are normalised before anything is counted. The make prefix is stripped from the trade name (TOYOTA YARIS CROSS → Yaris Cross), BMW is grouped by series (X1 XDRIVE25E → X1, 330E → 3 Series), Mercedes-Benz by class (GLC 300 E 4MATIC → GLC) and Volkswagen's electric range by model (ID.4 PRO 210KW → ID.4). Spelling variants are folded into one brand or model: TESLA MOTORS → Tesla, MERCEDES-AMG → Mercedes-Benz, CEE D → Ceed. Camper builders such as Adria, Hymer and Bürstner appear as their own merk in the register and keep their own brand rather than being folded into the chassis maker. Three models the register files under another model's trade name are told apart by their type-approval variant instead: RDW writes TOYOTA COROLLA for the Corolla Cross, TOYOTA C-HR for the electric C-HR+ and PANDA for the Grande Panda, and each gets its own row here. The full alias table is in the repository.

Checks

Every refresh runs the same gates, and any one of them stops the run before anything is published. A staleness gate rejects a dataset whose truth date is more than ten days old. An acceptance gate requires the top 40 makes to cover at least 97 % of registrations over the last twelve months and rejects same-prefix sibling makes and single-letter model names other than the Mercedes-Benz classes, Tesla's S, X and Y and Zeekr's X. A sanity gate rejects an export whose latest month is less than half the previous one, whose marts are empty, whose month series has a gap since January 2023, or that publishes a month with no freeze behind it. The CBS reconciliation gate compares each frozen month's new cars against 85898NED, at 1.5 % for a freshly frozen month, 3 % for a backfilled or still-provisional one and 5 % for a month frozen more than three years after it ended. A fuel gate keeps the other group under 2 % and a fuel-coverage gate keeps plates with no fuel row under 0.5 %. Finally a JSON audit re-derives every share, rank, month-over-month, year-over-year and year-to-date figure from the exported files and compares it with what was written.

Caveats

• Counts are cars first registered in the Netherlands in the month, measured on the RDW register snapshot taken a few days after the month ended and frozen there, so a month never changes once published. Cars exported later stay in the register for two years and only scrapped cars vanish: measured against CBS, a 36-month window loses at most 1.4 % (2025 100.6 %, 2024 99.6 %, 2023 98.6 %, 2022 97.6 % of the official new-car counts).
• Percentages are shown only when the comparison month has at least 30 registrations.
• Brands with fewer than 50 registrations over the 36 months have no brand page and appear as plain text in the tables.
• Registration is not sale: dealer demonstration stock and leasing fleets register cars too, and a used import is a car changing country rather than a car being sold new.

Data files and licence

Every page is generated from JSON under /data/nl/: meta.json, months.json, YYYY-MM.json (with kpis, models[], makes[], bev_models[], bev_makes[] and import_age[]), makes.json and makes/<slug>.json. The RDW data is CC0, the CBS table is CC BY 4.0, and this site's content and derived data are CC BY 4.0 — attribute "auto-stats.eu (data: RDW, CBS)".