How Long Does It Take to Crack a Password?
"Your password would take 3 billion years to crack" is a real statistic, but it rests on assumptions most tools don't show you. Here's what's actually behind the number.
The Basic Formula
Crack time is derived directly from entropy and an assumed guessing rate:
The "−1" reflects that, on average, an attacker finds the correct password after searching half the total space, not all of it. Everything in this formula except entropy is an assumption — and that assumption is where estimates diverge wildly.
Why the Same Password Shows Different Times on Different Sites
The guessing rate assumption varies enormously depending on the attack scenario:
| Scenario | Approx. guesses/second |
|---|---|
| Online attack against a rate-limited login form | 10-100 |
| Offline attack, slow salted hash (bcrypt, Argon2) | 10-10,000 |
| Offline attack, fast unsalted hash (MD5, SHA-1), single GPU | ~10 billion |
| Offline attack, fast hash, large GPU cluster | 100 billion+ |
A password with 60 bits of entropy could take a few minutes (against an unsalted fast hash and a GPU cluster) or several thousand years (against a rate-limited login form) — the entropy hasn't changed, only the assumption about how fast an attacker can guess.
What Crack-Time Estimates Don't Cover
- Credential stuffing — reusing a password leaked from another breach doesn't require "cracking" anything; the attacker already has it.
- Phishing — the strongest password in the world offers zero protection if you type it into a fake login page.
- Dictionary and pattern attacks — tools that try common passwords, leaked passwords, and known patterns first will find a weak-but-"high-entropy-looking" password (like
Password123!) in seconds, regardless of what the raw math suggests. - Rate limiting and account lockout — a system that locks an account after a handful of failed attempts makes online brute-force irrelevant no matter the password's theoretical crack time.
How to Read a Crack-Time Number Sensibly
Treat crack-time estimates as a rough comparative signal, not a literal guarantee — useful for comparing two candidate passwords against the same assumed attack model, not as a promise about your specific account's real-world safety. The safest interpretation of "this would take billions of years to crack" is: this password is not the weak link, not this password can never be compromised.
Frequently Asked Questions
They assume different guessing rates and attack models. Some assume a single GPU against an unsalted hash; others assume a rate-limited login form. The same password's entropy produces very different time estimates under each assumption.
For a conservative worst-case estimate, assume 10-100 billion guesses per second (an offline attack against a fast hash with modern GPU hardware). Well-designed systems using bcrypt or Argon2 reduce this to a few hundred or thousand guesses per second.
No — most real account compromises come from credential stuffing, phishing, or a leaked breach database, not brute-forcing a password's full character space. Crack-time math addresses one specific, comparatively rare attack scenario.
See a real crack-time estimate
Test any password with transparent, adjustable assumptions.
Use the Crack Time Calculator →