OpenStreetMap is the largest open geographic database in existence, but “download some of it” is famously non-obvious. The full planet file is over 80 GB, the export tab on openstreetmap.org caps out at a few thousand nodes, and Overpass QL has a learning curve that stops most people at the first curly brace.
What most GIS work actually needs is simpler: the buildings, roads or points of interest for one town or study area, in a format QGIS or a web map can open. Here are the practical routes, from easiest to most powerful.
1. Browser extraction for a drawn area
The fastest path for small-to-medium areas: draw a bounding box, pick a theme, and download OpenStreetMap data as GeoJSON, shapefile, KML or GeoPackage directly from your browser. The tool queries the Overpass API for you — no query language required — and converts the result client-side. Good for anything up to a city-sized extract of a single theme (all buildings, all highways, all amenities).
2. Geofabrik for country and region extracts
When you want everything for a country or region, Geofabrik’s daily extracts remain the standard. You get a .osm.pbf covering the area, which desktop GIS cannot open directly — plan on a conversion step. A browser-based OSM to shapefile converter handles the common case of pulling a layer out of a PBF without installing osmium or osm2pgsql.
3. Overpass Turbo for surgical queries
When you need “all fire hydrants within 500 m of a school” rather than a whole theme, Overpass Turbo is the tool — there’s no browser shortcut around learning a little Overpass QL for genuinely conditional queries. The export button gives you GeoJSON, which slots into any of the tools above.
4. Building basemaps: OSM to MBTiles
If the goal is a self-hosted basemap rather than analysis data, you want vector tiles, not features. Converting an extract with an OSM to MBTiles converter produces a single tileset file you can serve offline or from object storage — increasingly common for field apps that can’t rely on connectivity.
Licensing, briefly
OSM data is ODbL-licensed: attribution (“© OpenStreetMap contributors”) is mandatory, and share-alike applies to derived databases (not to maps you render from the data). For most analysis and cartography this is a footnote, but it matters if you’re merging OSM into a proprietary dataset.
For administrative boundaries specifically — country, state and district outlines — OSM is usually not the best source; dedicated boundary datasets are cleaner. See our guide to downloading free administrative boundary data.
