GUID vs UUID: The Complete Developer Guide to Unique Identifiers

GUID and UUID are the same thing — Globally Unique Identifier and Universally Unique Identifier — just different naming conventions. Microsoft historically used "GUID" in COM/COM+ and .NET, while the broader industry standardized on "UUID" via RFC 4122. Under the hood they're both 128-bit identifiers formatted as xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx.

GID

Open in Dev Cosmos

GUID Generator →

↗

Anatomy of a UUID

A UUID has five groups separated by hyphens:

Copy550e8400-e29b-41d4-a716-446655440000
|______| |___| |___| |___| |__________|
8 chars  4     4     4     12
time_low time  ver+  var+  node
         mid   hi    clock

The 4-bit version field (characters 13–13) encodes which UUID version was used to generate it. The 2-bit variant field signals the layout follows RFC 4122.

UUID Versions

v1 — Time-Based

Version 1 derives uniqueness from the current timestamp (measured in 100-nanosecond intervals since October 15, 1582) plus the generating machine's MAC address. The result is sortable by creation time, which is an advantage for database indexes.

v4 — Random (Most Common)

Version 4 uses 122 random bits from a cryptographically secure RNG. No coordination, no infrastructure, no state. It's the go-to choice for most applications:

Copy// JavaScript
crypto.randomUUID() // → "550e8400-e29b-41d4-a716-446655440000"

// Python
import uuid; str(uuid.uuid4())

// C#
Guid.NewGuid().ToString()

v5 — Namespace + Name (Deterministic)

Version 5 is deterministic: the same namespace UUID + name string always produces the same UUID. It uses SHA-1 internally. This is useful when you need stable IDs from content:

• Converting a URL into a stable UUID for deduplication
• Generating user IDs from email addresses consistently across services
• Creating reproducible test data IDs

Nil UUID

All 128 bits set to zero: 00000000-0000-0000-0000-000000000000. Used as a sentinel "not set" value in APIs and databases, similar to null.

Collision Probability

For v4 (random), the probability of generating a duplicate when creating 1 billion UUIDs per second for 85 years is approximately 50%. In practice, the chance of collision in any realistic application is effectively zero.

UUIDs Generated	Collision Probability
1,000	~0.000000000000000000001%
1 billion	~0.00000000000000001%
10 trillion	~0.0000001%

Output Formats

Format	Example	Use Case
Standard	550e8400-e29b-41d4-a716-446655440000	Default, most APIs
UPPERCASE	550E8400-E29B-41D4-A716-446655440000	SQL Server, legacy .NET
Braces	{550e8400-e29b-41d4-a716-446655440000}	.NET GUIDs, C++ COM interfaces
No hyphens	550e8400e29b41d4a716446655440000	Compact storage, URL path segments

UUIDs as Database Primary Keys

Using UUIDs as primary keys has trade-offs. The main advantage is that keys can be generated on the client without a database round-trip, enabling offline-first architectures. The main disadvantage is that random v4 UUIDs cause B-tree index fragmentation, degrading insert performance at scale.