Why are my three credit scores different?

You pull three scores and get three numbers. The instinct is that two of them must be wrong. Usually none of them are — you're looking at three different calculations run on three different data sets, and both of those differences are doing work.

Two causes, stacked

Almost every confusing explanation of this treats it as one problem. It's two, and they compound:

1. Your three credit reports don't contain the same information.
2. Even where they do, they're scored by different algorithms.

Separating these tells you whether your gap is something to look into or something to ignore.

Cause one: the three reports hold different data

The foundational fact, and the one most people have never been told: reporting to the credit bureaus is entirely voluntary. No law requires any creditor to report to any bureau, let alone all three. The CFPB states this plainly.

Once you know that, the rest follows:

• Not every creditor reports to all three. Each bureau maintains its own subscriber relationships. A card you've held for eight years might appear on two reports and not the third — which means one of your files is missing eight years of positive history the other two have.
• Update timing differs. Furnishers report on their own schedules, usually tied to statement dates. The same card can show three different balances simply because each bureau captured a different day of the month. On a card carrying a balance, that alone can produce a meaningful score difference through utilization.
• Each bureau assembles your file with its own matching logic. That's where mixed files come from — someone else's tradelines attached to your record — and split files, where your own history gets divided across two partial records.
• Errors are rarely symmetrical. A collection reported to one bureau but not the others is common. So is an inquiry, or a wrong balance.

Cause two: different scoring models

Even on identical data, different models return different numbers. And there are far more models in active use than most people realize.

FICO and VantageScore are different companies' products. They're both on a 300–850 scale, which invites the assumption that they're comparable. They aren't. They weight things differently and handle thin files differently.

There are many FICO versions running at once. FICO 8 is the most widely used general model. FICO 9 and FICO 10 exist. And mortgage lending runs on much older models — FICO 2, 4, and 5. A lender chooses which version to license, which is why two lenders can pull the same bureau on the same day and quote you different numbers.

Free apps usually show an educational score. Often a VantageScore, sometimes a FICO 8. It's a real score computed from real data. It's just frequently not the one anyone will lend against.

What's a normal spread?

A spread of roughly 20 to 40 points across three bureaus, using the same model, is ordinary. I see it constantly on files with nothing wrong.

That range comes from exactly the mechanics above: one card reported to two bureaus and not the third, one balance captured a week later than another, one old account that aged off one file before the others.

What I'd treat as ordinary noise: a gap under about 40 points with no obvious cause. What I'd treat as a question worth answering: a gap materially larger than that, or a gap that appeared suddenly.

When a gap is worth investigating

Four patterns that make me look harder:

• One bureau is 50+ points below the other two. That's rarely a modeling artifact. Usually there's a specific derogatory item on that one report — a collection, a late, or a maxed-out card with a missing credit limit.
• A score dropped sharply on one bureau only. Something new landed on that file. Pull it and find out what.
• One report shows accounts you don't recognize. Check the personal information section on that report immediately. Unfamiliar names or addresses alongside unfamiliar accounts is the signature of a mixed file or identity theft, and it's a file-level problem, not an account-level one.
• One report is unusually thin. If your oldest accounts appear on two reports and not the third, that third file is being scored as a shorter history than you actually have.

You're entitled to see all three. FCRA § 612(a) (15 U.S.C. § 1681j(a)) guarantees a free file disclosure from each nationwide bureau once every twelve months, and the three bureaus currently offer free weekly access on top of that — a voluntary industry program rather than a statutory right, but it's in effect. Pull all three. Comparing one report against another is how you find asymmetric errors, and it's not something you can do from a single score.

Which score actually counts

If a mortgage is anywhere in your plans, most of the above becomes secondary to one specific arrangement.

Your lender orders a tri-merge report — all three bureaus in one document, scored with the classic mortgage models:

• FICO Score 2 — Experian
• FICO Score 4 — TransUnion
• FICO Score 5 — Equifax

Then the lender uses your middle score. Not the average, not the highest. With a co-borrower, each of you gets a middle score and the loan is qualified on the lower of those two.

The strategic consequence is one most people miss: raising your highest score accomplishes nothing. If your three are 640, 668, and 702, your qualifying number is 668. Work that moves the 702 is wasted. Work that moves the 640 only matters if it pushes past 668 and changes which score sits in the middle. Everything should aim at the middle-score bureau.

Expect these classic models to read lower than your app score — they're older and less forgiving. Full comparison of mortgage vs. consumer reports.

What to actually do about it

Pull all three reports and read them side by side. Any item on one report and not the others is either a legitimate difference in what got furnished, or an error — and you can only tell which by comparing. Start with how to read a credit report section by section if you want the field-by-field walkthrough.

Then check credit limits on every revolving account, on every report. A blank limit is one of the most common asymmetric errors and one of the most costly, because it distorts how your utilization reads on that bureau alone. Each bureau labels these fields differently — if the odd one out is TransUnion, here's how to read a TransUnion credit report field by field.

Dispute bureau by bureau. There's no shared correction mechanism — fixing an item at one bureau does nothing at the other two, and for mortgage purposes a fix at the wrong bureau doesn't change your qualifying score at all.