China Administrative BoundariesGB/T 2260 · WGS 84 Browse datasets
Downloadable vector data

China administrative boundaries, in the formats you use.

34 provinces, 368 prefectures and 2,890 counties. Each unit carries its GB/T 2260 code and its Chinese name next to the English one, and comes as GeoJSON, shapefile, SVG, DXF, GML, KML or CSV. Buy the whole country, or one province at a time.

34Provinces
368Cities & prefectures
2,890Counties & districts
7File formats
All 34 province-level divisions
All 34 province-level divisionsWGS 84 · EPSG:4326
The datasets

Three levels, one set of fields.

Buy the country or a single province. The fields are the same at every level, so a provincial file drops into a national build without renaming a column. Admin 1 is the province, Admin 2 the prefecture, Admin 3 the county, which is the numbering GADM and geoBoundaries use.

Level 1 34 units

China Province Boundaries

$19.99one-time
  • 34 province-level divisions
  • Maritime boundary lines included
  • GeoJSON, SHP, SVG, DXF, GML, KML, CSV
Level 3 2,890 units

China County & District Boundaries

$99.99one-time
  • 2,890 counties & districts
  • Parent province and city on every row
  • GeoJSON, SHP, SVG, DXF, GML, KML, CSV
What you get

What arrives in the zip.

GeoJSON for code, shapefile for desktop GIS, SVG and DXF for design, GML, KML, and CSV if you only want the table. Same polygons in every one of them.

Formats & schema

Seven formats, one download

GeoJSON.geojson

RFC 7946, the reference version of the data.

Code, web maps, PostGISWGS 84
Shapefile.shp

With .shx, .dbf, .prj and .cpg. UTF-8 attributes intact.

ArcGIS, QGIS projectsWGS 84
SVG.svg

One group per division, keyed by its code. Names sit in a layer you can hide.

PowerPoint, Illustrator, InkscapeDegrees
DXF.dxf

R12 ASCII, projected to metres, simplified to about 45 m.

AutoCAD, illustrationMetres, LCC
GML.gml

OGC GML 3.2, latitude written before longitude.

GeoServer, XML pipelinesWGS 84
KML.kml

Styled so it looks reasonable when you double-click it.

Google Earth, quick checksWGS 84
CSV.csv

UTF-8 with BOM, so Excel shows the Chinese names instead of squares.

Excel, pandas, joins—
FormatsGeoJSON · SHP · SVG · DXF · GML · KML · CSV, one zip
Coordinate systemWGS 84 · EPSG:4326 · decimal degrees
GeometryMultiPolygon — one feature per division
Codes6-digit GB/T 2260, e.g. 440300
Namesname_en and name_zh on every feature
Hierarchyprovince_code → city_code → code
Extrasparts, center_lon, center_lat, bbox
LicenceCommercial use, one organisation, no attribution required
DeliveryInstant download; links re-issued whenever you open them
Data current to

The parts that eat a week.

01

The code is already there

Every feature carries its six-digit GB/T 2260 code, so it joins to National Bureau of Statistics tables without matching names by hand.

02

Both names, on every feature

name_en and name_zh sit side by side. One file labels a map in either language, and there is no lookup table to maintain at render time.

03

Seven formats, one geometry

GeoJSON, shapefile, SVG, DXF, GML, KML and CSV, all cut from the same polygons. The CAD file and the dataframe come from the same source.

04

Opens where you work

WGS 84 throughout, except the DXF, which is projected to metres so CAD behaves. Drag it into QGIS, ogr2ogr it into PostGIS, read it into pandas.

Open it before you buy.

Zhejiang's city layer, free and without a signup: eleven features, the full field set, all seven formats. It is the same data you would be buying.

Questions

All questions
What exactly do I get?

One zip: the dataset in seven formats, a README describing every field, and the licence. One payment, one download link, no subscription.

Which format should I use?

GeoJSON if you're writing code or using Mapbox or Leaflet. Shapefile for ArcGIS and QGIS projects, DXF for AutoCAD and Illustrator, GML for GeoServer, KML for Google Earth, CSV if you only want the table. They are all in the same zip, so there's nothing to get wrong.

Is the data up to date?

The division list follows current GB/T 2260 codes, including the county-level changes of recent years. Each file records the date it was generated.

Can I use it commercially?

Yes. One purchase covers one organisation, client work included. The one thing you can't do is resell the data as a dataset.