Suggest an editImprove this articleRefine the answer for “What is a transaction isolation level?”. Your changes go to moderation before they’re published.Approval requiredContentWhat you’re changing🇺🇸EN🇺🇦UAPreviewTitle (EN)Short answer (EN)A **transaction isolation level** is a setting that determines **how much one transaction can see of changes made by others** while those are still uncommitted; the higher the level, the fewer conflicts, but the lower the performance. **Key point:** the SQL standard defines 4 levels - `Read Uncommitted` (can see even uncommitted changes from others), `Read Committed` (sees only committed data), `Repeatable Read` (the same read within a transaction returns the same data, though phantom rows are possible), and `Serializable` (maximum isolation, transactions behave as if run sequentially).Shown above the full answer for quick recall.Answer (EN)ImageA **transaction isolation level** is a setting that determines **how much one transaction can see of changes made by others** while those are still uncommitted. In simpler terms, it's the **degree of "separation"** between concurrent transactions. The higher the level, the fewer conflicts, but the lower the performance. ### The 4 isolation levels The SQL standard defines 4 levels: 1. **Read Uncommitted**, you can see even *uncommitted* changes from other transactions (risk of a "dirty read"). 2. **Read Committed**, you only see *committed* data, but it can change between two reads. 3. **Repeatable Read**, the same read within a transaction returns identical data, though "phantom rows" are possible (new records added by another transaction). 4. **Serializable**, maximum isolation; transactions behave as if they ran sequentially. Safe, but slow. So the isolation level tunes the balance between **data accuracy** and system **performance**.For the reviewerNote to the moderator (optional)Visible only to the moderator. Helps review go faster.