The numbers,
with sources.
Compiled from central and state notifications. Every figure on this page links to the document it came from — including the ones that complicate our own argument.
- Central government quota
- 59.5%
- Judicially settled ceiling
- 50%
- States legislating above it
- 8+
- Amendment — added 10% EWS
- 103rd
Central allocation
How the 59.5% at the central level is currently apportioned, with open merit as the remainder.
- OBC27% · Mandal, 1990
- SC15% · since 1950
- EWS10% · 103rd Amendment
- ST7.5% · since 1950
- Open merit40.5% · unreserved
Source · DoPT consolidated instructions · Open the document →
State explorer
Select a region to see its reported total. Several of these are under challenge — we say so rather than hiding it.
Against the 50% ceiling
Red = above the 50% ceilingFigures as notified by each government; several are sub judice
View as a table
| Region | Total | As of | Source |
|---|---|---|---|
| Central Government | 59.5% | 2026-07-01 | RHA-2025-0354 |
| Tamil Nadu | 69% | 2026-07-01 | RHA-2025-0367 |
| Jharkhand | 77% | 2026-07-01 | RHA-2025-0367 |
| Chhattisgarh | 72% | 2026-07-01 | RHA-2025-0367 |
| Bihar | 65% | 2026-07-01 | RHA-2025-0367 |
| Rajasthan | 64% | 2026-07-01 | RHA-2025-0367 |
| Madhya Pradesh | 63% | 2026-07-01 | RHA-2025-0367 |
| Haryana | 60% | 2026-07-01 | RHA-2025-0367 |
| Gujarat | 59% | 2026-07-01 | RHA-2025-0367 |
Since 1950
How the central ceiling has moved. Each step is a document in the repository.
How this page is compiled
Figures are transcribed from the notification or judgment that created them, never from secondary reporting. Where a state figure is contested in court, we mark it and link the pleading.
Percentages are of advertised posts or seats, not of population. Comparing the two is the single most common misreading of this data.
The gaps in the record
- Sub-category-wise outcomes over time, for most departments.
- Consistent creamy-layer exclusion figures across states.
- Backlog vacancy data published on a comparable basis.
Download
Take the dataset
Everything on this page as CSV, with a source column pointing at the document ID in the repository.