How to Download LSOA Data: Complete Guide

What is an LSOA?

A Lower Super Output Area is a small statistical geography used by the Office for National Statistics (ONS) to publish data at neighbourhood level. There are 33,755 LSOAs in England and Wales (2021 boundaries), each containing approximately 1,000 to 3,000 residents, with a target of 1,500.

LSOAs were designed to be stable over time, similar in population size, and to nest neatly within local authority boundaries. Unlike wards (which are political and change at every boundary review), LSOAs change only when population shifts make them too large or too small — and even then, changes are minimal. Between the 2011 and 2021 censuses, approximately 97% of LSOAs remained unchanged.

This stability is what makes LSOAs the preferred geography for time-series analysis, deprivation indices, and any work that requires consistent small-area comparisons.

The LSOA Code Format

Every LSOA has a unique 9-character code:

• English LSOAs: start with E01 (e.g., E01000001)
• Welsh LSOAs: start with W01 (e.g., W01000001)

The codes are assigned by ONS and are used as the primary key across all government statistical datasets. When you are joining census data to deprivation data, health data, or education data at LSOA level, this code is your link.

Method 1: ONS Nomis (Interactive)

Nomis (nomis.co.uk) is the official ONS labour market and census statistics portal. It is the most accessible way to download LSOA-level census data without writing code.

Step-by-Step

1. Go to nomisweb.co.uk and click "Census 2021" in the navigation
2. Browse the available table series (TS for topic summaries, RM for detailed cross-tabulations)
3. Select your table — e.g., TS001 (Number of usual residents)
4. Under "Geography," select "2021 Lower Layer Super Output Areas" as the geographic level
5. Choose your area filter — you can select all LSOAs, or filter by region, local authority, or specific LSOA codes
6. Select your output format (CSV is best for data analysis; Excel for quick viewing)
7. Click "Download data"

What You Get

A CSV file with one row per LSOA and columns for each variable in the table. File sizes range from a few MB (single variable, single region) to 100+ MB (full cross-tabulation, all LSOAs).

Limitations

Nomis has a download limit of approximately 2 million cells per query. For large cross-tabulations across all LSOAs, you may need to split your download by region or variable group. The interface can also be slow for very large selections.

Method 2: ONS Census Data Downloads (Bulk)

For large-scale analysis, the ONS bulk download page is more efficient than Nomis.

Where to Find It

Go to ons.gov.uk/census and navigate to "Data and analysis from Census 2021." ONS publishes complete table downloads as zipped CSV files, one per table and geographic level.

File Structure

Each bulk download typically contains:

• A data file (CSV) with geography codes as rows and variable codes as columns
• A metadata file explaining variable codes and category labels
• A geography lookup mapping codes to area names

The LSOA-level files are the largest. For example, the population by age table at LSOA level contains 33,755 rows and 90+ columns — manageable in any data tool but too large for Excel's point-and-click comfort zone.

Recommended Tables for LSOA Analysis

Table Code	Topic	Key Variables
TS001	Usual residents	Total population count
TS007	Age by single year	Age 0 to 100+
TS009	Sex	Male, female
TS021	Ethnic group	19 ethnic group categories
TS037	General health	Very good to very bad (5 categories)
TS054	Tenure	Owned, shared, social rented, private rented
TS066	Economic activity	Employed, self-employed, unemployed, inactive
TS067	Highest qualification	No qualifications to Level 4+

Method 3: ONS Census API (Programmatic)

ONS provides an API for the 2021 Census data that allows programmatic queries. This is the best option for developers building applications or analysts who need to automate data retrieval.

Base URL

The Census 2021 API is accessed via: https://api.beta.ons.gov.uk/v1/

Key Endpoints

• Datasets: list all available census datasets
• Observations: retrieve data for specific geographies and variables
• Geography: browse the geographic hierarchy

Example Query

To get population data for a specific LSOA, you would construct a query specifying the dataset ID, the geography code, and the variable dimension. The API returns JSON responses.

Rate Limits

The API has usage limits that are appropriate for application development but not for bulk downloads. If you need all LSOAs, use the bulk CSV downloads instead.

Method 4: ONS Open Geography Portal (Boundaries)

If you need to map LSOA data or perform spatial analysis, you need the boundary files as well as the statistical data.

Boundary Downloads

The ONS Open Geography Portal (geoportal.statistics.gov.uk) provides:

• LSOA boundary shapefiles (GeoJSON, GeoPackage, or Shapefile format)
• LSOA centroids (population-weighted point coordinates)
• Lookup tables mapping LSOAs to higher geographies (MSOAs, wards, local authorities, regions)

Practical Workflow

1. Download LSOA statistical data from Nomis or ONS bulk downloads
2. Download LSOA boundary files from the Open Geography Portal
3. Join the two on the LSOA code
4. Visualise in QGIS (free), Python (geopandas), R (sf package), or any GIS tool

Method 5: Deprivation Data at LSOA Level

The English Indices of Deprivation (IMD) are published at LSOA level. This is one of the most commonly requested LSOA datasets.

Download Location

Go to gov.uk/government/statistics/english-indices-of-deprivation-2019 and download "File 1: Index of Multiple Deprivation." The file contains:

• Overall IMD rank and score
• Decile assignment (1 = most deprived 10%)
• Seven domain-level ranks and scores
• IDACI and IDAOPI sub-domain scores

Joining IMD to Census Data

Since both datasets use LSOA codes as the primary key, joining is straightforward. However, note that IMD 2019 uses 2011 LSOA boundaries while Census 2021 uses 2021 boundaries. Use the ONS boundary change lookup to reconcile the ~3% of LSOAs that changed between versions.

Working with LSOA Data: Practical Tips

File Size Management

Full LSOA-level census downloads can exceed 100 MB. For efficient processing:

• Use parquet format (convert from CSV) for fast column-oriented reads in Python or R
• Filter to your area of interest before loading the full dataset
• Use DuckDB or SQLite for SQL queries on large CSVs without loading them into memory

Aggregation

LSOA data nests neatly into higher geographies. To aggregate to MSOA level, use the LSOA-to-MSOA lookup from the ONS Geography Portal. To aggregate to local authority level, use the LSOA-to-LAD lookup. Always aggregate using population-weighted methods when calculating rates or percentages.

Common Joins

The most frequent data joins at LSOA level:

Dataset A	Dataset B	Join Key
Census 2021 population	IMD 2019 deprivation	LSOA code (with 2011→2021 lookup)
Census 2021 data	Postcode list	LSOA code (via NSPL postcode lookup)
Census 2021 data	LSOA boundaries	LSOA code
Census 2021 LSOA	Census 2011 LSOA	LSOA code (with change lookup)
Census 2021 data	Crime data (police.uk)	LSOA code

Disclosure Control

ONS applies statistical disclosure control (SDC) to LSOA-level census data. Small counts (typically below 3) may be rounded or suppressed. If you are analysing rare characteristics in small areas, check whether the counts you see are real or suppressed. The ONS methodology guides for each table explain the SDC approach used.

LSOA Data for Scotland and Northern Ireland

Scotland uses a different small-area geography called Data Zones (equivalent to LSOAs). There are approximately 6,976 Data Zones in Scotland. Scotland's Census 2022 data at Data Zone level is published by National Records of Scotland at scotlandscensus.gov.uk.

Northern Ireland uses Super Output Areas (SOAs), of which there are 890. NISRA publishes Northern Ireland census data at SOA level via ninis2.nisra.gov.uk.

The geographies are not directly comparable across the four UK nations, but the population sizes are similar (1,000–3,000 residents per unit), so analytical methods transfer well.

Frequently Asked Questions

---

Related Guides