Census Output Areas Explained: OA, LSOA, MSOA

The Census Geography Hierarchy

The Office for National Statistics (ONS) publishes census data at multiple geographic levels. Each level nests within the one above it, like a set of containers:

Level	Count (2021)	Typical Population	Code Prefix
Output Area (OA)	~188,880	100–625 (target ~300)	E00 / W00
Lower Super Output Area (LSOA)	33,755	1,000–3,000 (target ~1,500)	E01 / W01
Middle Super Output Area (MSOA)	7,264	5,000–15,000 (target ~7,200)	E02 / W02
Local Authority District (LAD)	331	2,000–1,145,000	E06–E09 / W06
Region	10	3.2M–9.3M	E12 / W92
Country	2	3.1M–56.5M	E92 / W92

Each Output Area nests within exactly one LSOA. Each LSOA nests within exactly one MSOA. Each MSOA nests within exactly one local authority. This clean nesting means you can aggregate data upward from OA to LSOA to MSOA to LAD without any overlap or double-counting.

Output Areas (OAs)

Output Areas are the smallest geography at which census data is published. They were first created for the 2001 Census and are designed to be:

• Roughly equal in population — target of ~300 people (40–250 households)
• Socially homogeneous — OAs try to group together households with similar tenure type (owned, rented, social) and dwelling type (detached, semi, flat)
• Compact in shape — ideally contiguous and compact, avoiding elongated or fragmented shapes
• Stable over time — OAs change only when population growth or decline makes them too large or too small

When to Use OAs

OA-level data is the most granular available, but it comes with significant limitations:

Advantages:

• Street-level resolution — an OA covers roughly one block of houses or one apartment building
• Essential for spatial analysis requiring fine granularity (e.g., overlaying census data with flood risk zones)
• Foundation for building custom catchment areas

Disadvantages:

• Statistical disclosure control: ONS rounds or suppresses small counts at OA level to prevent identification of individuals. This means many cells in OA-level tables contain rounded values (e.g., 0, 3, 6, 9) rather than exact counts
• Small numbers, large volatility: With only ~300 people per OA, percentages swing wildly. An OA with 3 people aged 85+ out of 300 shows "1.0%"; the next OA might show "2.3%". These differences are statistical noise, not meaningful variation
• Large file sizes: 188,880 rows per table, multiplied by dozens of variables, creates datasets that are unwieldy for most tools

For most practical applications, LSOA-level data is the sweet spot — granular enough for neighbourhood analysis but with large enough populations to produce reliable statistics.

Lower Super Output Areas (LSOAs)

LSOAs are the workhorse geography of UK small-area statistics. Each LSOA is built from a cluster of 4–6 contiguous Output Areas and has a target population of approximately 1,500 people.

Why LSOAs Matter

LSOAs are the primary geography for:

• The Index of Multiple Deprivation (IMD) — the most widely used deprivation measure in England
• Health data — Public Health England profiles, NHS service planning
• Crime data — police.uk publishes crime statistics at LSOA level
• Education data — school-level deprivation funding uses LSOA-level IDACI
• Census cross-tabulations — many detailed cross-tabulations are only available at LSOA level and above (not at OA level)

LSOA Names

Since the 2021 Census, ONS has assigned names to MSOAs (e.g., "Tower Hamlets — Bethnal Green South"). LSOAs do not have official names — they have codes (e.g., E01000001) and are typically identified by their parent MSOA name plus a numeric suffix. Third-party tools sometimes generate LSOA names by combining the parish or ward name with a directional indicator (North, South, etc.).

2011 vs. 2021 Boundaries

The 2021 Census introduced a revised set of LSOA boundaries. Most LSOAs (approximately 97%) are unchanged from 2011, but some were split (where population grew too large) or merged (where population shrank). ONS publishes a boundary change lookup table that maps 2011 LSOA codes to their 2021 equivalents.

This is particularly important for time-series analysis: if you want to compare 2021 Census data with IMD 2019 data (which uses 2011 boundaries), you need the lookup table to align the geographies.

Middle Super Output Areas (MSOAs)

MSOAs are built from clusters of 4–5 LSOAs and have a target population of approximately 7,200 people. There are 7,264 MSOAs in England and Wales (2021 boundaries).

MSOA Names

One significant improvement in the 2021 Census geography is that MSOAs now have official names assigned by ONS, developed in consultation with local authorities. These names are based on local place names and landmarks — e.g., "Manchester — Northern Quarter", "Birmingham — Selly Oak", "Cardiff — Canton".

Previously, MSOAs were identified only by codes (e.g., E02000001), which made them inaccessible to non-specialist audiences. The named MSOAs make census data far more useful in public-facing reports and dashboards.

When to Use MSOAs

MSOAs are the best choice when:

• Creating maps: 7,264 units produces readable choropleth maps. 33,755 LSOAs can look like noise at national scale; 188,880 OAs are unwieldy
• Publishing results for non-technical audiences: the named MSOAs ("Islington — Highbury East") are immediately understandable
• Analysing detailed cross-tabulations: some census tables are only published at MSOA level and above to protect statistical confidentiality
• Aggregating for dashboards: MSOA is the right grain for neighbourhood-level performance indicators in public health, policing, and local government

How Output Areas Are Created

ONS does not draw output areas by hand. They are generated algorithmically:

1. Start with postcodes: each postcode is assigned grid reference coordinates
2. Cluster postcodes: an automated algorithm groups postcodes into OAs that meet the population thresholds and social homogeneity criteria
3. Constrain to boundaries: OAs must nest within local authority and ward boundaries
4. Manual review: ONS geographers review and adjust boundaries where the algorithm produces odd shapes or splits communities

The algorithm prioritises population size (staying within the target range), then shape compactness, then social homogeneity. In practice, OAs in urban areas are geographically tiny (a few streets) while OAs in rural areas can cover many square kilometres.

The Frozen Boundary Problem

Output areas are "frozen" between censuses — they do not change from year to year. This means that between census years, some OAs in rapidly growing areas may contain far more than 625 people, while depopulating areas may have OAs with fewer than 100 people. ONS addresses this by redrawing boundaries for each census, but this creates discontinuities that complicate time-series analysis.

How Output Areas Relate to Other Geographies

Wards

Electoral wards are the political geography used for council elections. They vary widely in size (from ~5,000 to ~20,000 people) and change frequently at boundary reviews. Output Areas nest within wards, but because ward boundaries change more often than OA boundaries, the nesting relationship can break at boundary revisions.

For census purposes: use output areas (OA/LSOA/MSOA) for statistical analysis, not wards. Wards are political units with no statistical design criteria.

Postcodes

Postcodes do not nest within output areas — a single postcode can straddle an OA boundary. The ONS National Statistics Postcode Lookup (NSPL) assigns each postcode to the OA that contains its population-weighted centroid. This mapping is good enough for most practical purposes but is not exact.

Parliamentary Constituencies

Parliamentary constituencies are built from whole wards, not from output areas. However, ONS publishes lookup tables that map OAs and LSOAs to constituencies (using a "best fit" approach for OAs that straddle constituency boundaries). This enables census data analysis at constituency level — useful for political research and policy analysis.

Clinical Commissioning Groups / Integrated Care Boards

Health geographies use a "best fit" mapping from LSOAs. Each LSOA is assigned to the health geography that contains the majority of its population. This means you can analyse census demographics for any health administrative area by aggregating the LSOA data.

Practical Workflow: Choosing the Right Level

Use Case	Recommended Level	Reason
National overview maps	MSOA or LAD	Readable maps, named areas
Deprivation analysis	LSOA	IMD is published at LSOA level
Postcode demographic lookup	LSOA	NSPL maps postcodes to LSOAs
Retail site selection	LSOA or OA	Need street-level granularity
School catchment profiling	LSOA	Balance of granularity and data quality
Time-series comparison (2011 vs. 2021)	LSOA (with lookup)	97% unchanged; manageable boundary changes
Cross-tabulation analysis	MSOA or LAD	Detailed tables may not be available at LSOA
Custom area aggregation	OA	Finest building block for creating bespoke areas

Data Downloads

All output area boundary files and lookup tables are available free from the ONS Open Geography Portal (geoportal.statistics.gov.uk):

• Boundary files: GeoJSON, Shapefile, and GeoPackage formats for OA, LSOA, and MSOA
• Centroids: Population-weighted centroid coordinates for each area
• Lookup tables: OA-to-LSOA, LSOA-to-MSOA, MSOA-to-LAD, LSOA-to-ward, postcode-to-OA, and cross-boundary "best fit" lookups
• Names: MSOA official names (CSV download)
• Change histories: 2011-to-2021 boundary change lookups

All data is published under the Open Government Licence v3.0.

Frequently Asked Questions

---

Related Guides