Almost every piece of car-buying advice written in the last twenty years contains the same sentence: try a credit union, they are usually cheaper. It is repeated so often that it has stopped functioning as a claim and started functioning as furniture. Nobody says how much cheaper, on what product, measured how, or whether the difference has held still.

It is checkable, and the check is more interesting than the folklore. The National Credit Union Administration publishes a quarterly table comparing what credit unions and banks charge across a fixed list of products, and two of the rows are used-car loans. That single table settles the direction of the claim, puts a number on it for a stated date, and — read across several years rather than one quarter — shows that the number moves far more than the folklore admits.

This page is about the credit union channel specifically: what the institution is, how you become eligible to borrow from one, what the published comparison actually measures, and which of the rules that make credit unions distinctive apply to the one you are standing in front of. The general mechanics of a car loan — how a rate is assembled, why the advertised figure falls as the term lengthens, what a dealer’s margin does to a quote — are set out separately in our page on used-car financing rates, and are not repeated here.

The one federal table that prints both columns

The NCUA is an independent federal agency, created by Congress in 1970, which charters and regulates federal credit unions and insures deposits at federally insured ones. Four times a year it publishes a comparison of national average rates for 23 common loan and deposit products at banks and at credit unions. The agency does not survey the rates itself: each page states that NCUA extracted the figures from S&P Global Market Intelligence, and that they represent rates reported by active institutions for the last Friday of that quarter.

That provenance matters and is easy to skate over. This is a federal agency’s published table built on a commercial vendor’s survey of posted rates. It is not a regulatory filing, and it is not a census of loans actually written. Read it as what institutions were advertising on a particular Friday.

The Friday in question, for the most recent quarter available when this page was checked on 7 September 2026, was 26 December 2025. Four of the 23 rows are car loans, and they read like this.

On a used car over 48 months, the credit union average was 5.53% and the bank average 7.73% — a spread of 2.20 percentage points. On a used car over 36 months, 5.41% against 7.69%, a spread of 2.28. On a new car over 60 months, 5.44% against 7.41%, a spread of 1.97. On a new car over 48 months, 5.32% against 7.33%, a spread of 2.01.

So the folklore is directionally right on this reading, and the used-car rows carry the widest gaps of the four. What the folklore never mentions is the rest of the table. Of the 23 products compared, 21 favoured credit unions on that date — higher on a deposit, lower on a loan. Two did not, and both were the accounts an ordinary member uses every day: interest checking, where the bank average was 0.20% against the credit union average of 0.15%, and regular savings, 0.32% against 0.19%. The credit card row went the other way emphatically, 12.58% against 15.27%, as did the unsecured 36-month loan at 10.64% against 12.00%.

A channel that wins twenty-one rows and loses two is a strong result and a specific one. It is not a rule of nature, and the two rows it loses are the ones a member touches most often.

What a credit union is, and why the ownership shows up in the rate

The rate gap is downstream of a corporate structure, so it is worth being precise about the structure rather than gesturing at it.

NCUA’s own consumer material puts it in three sentences. Credit unions are owned and controlled by their members. A volunteer board of directors, elected by the members, manages the institution. All credit unions are not-for-profit organisations, and profits are returned to members in the form of reduced fees, higher savings rates and lower loan rates.

Read that last clause as a description of where the money goes rather than as a promise about your quote. A bank’s surplus has a claimant outside the customer base. A credit union’s does not, so the surplus can be routed back into pricing — but routing it into loan rates is a choice among three named destinations, and the same sentence lists cheaper fees and better savings rates as the alternatives. An institution that has decided to compete on deposit rates this year is a worse loan quote than one that has decided to compete on lending, and both are behaving exactly as the structure allows. That is one reason the published averages have moved as much as they have.

The other half of the definition is the safety net. Deposits are insured by the National Credit Union Share Insurance Fund, which NCUA manages. Backed by the full faith and credit of the United States, it covers accounts at all federal credit unions and the vast majority of state-chartered ones. The standard share insurance amount is $250,000 per share owner, per insured credit union, for each account ownership category.

None of that insures a borrower against anything — share insurance protects money you put in, not money you take out — but it is the reason a small institution with a name you have never heard of is not a risk you are taking on by moving your loan there.

The scale is larger than the word “union” suggests to people who have never used one. As of 31 March 2026, NCUA’s Call Report summary counted 4,250 federally insured credit unions holding $1.73 trillion in loans for 145.8 million members. Of that loan book, $479.6 billion was vehicle lending, and $321.3 billion of it — 67.0% of the vehicle book, 18.6% of every dollar these institutions have lent to anybody for any purpose — was used cars. New-car paper accounted for $158.3 billion.

Those two numbers moved in opposite directions over the year to that date: used-car balances grew 1.0% while new-car balances fell 2.2%, leaving the vehicle book as a whole essentially flat at a decline of 0.1%. Whatever else is true of this channel, used cars are not a sideline in it. They are the largest single vehicle category on its balance sheet by a wide margin.

The institutions themselves are consolidating, and quickly. There were 5,099 federally insured credit unions at the end of 2020 and 4,250 at the end of the first quarter of 2026, a fall of 161 in the last year of that span alone. The practical consequence for a borrower is mundane and real: the small employer-based credit union somebody recommends to you may have merged into a larger one with different pricing and a different field of membership since they last used it.

Membership is an eligibility test, not a credit test

The thing that stops most people trying this channel is a belief that they are not allowed in. It is worth separating that into two questions, because they have different answers.

Members of a credit union share a common bond, which the institution defines as its field of membership. NCUA lists four ordinary routes in: your employer, since many employers sponsor their own credit union; your family, since most credit unions let members’ relatives join; geography, since many serve anyone who lives, works, worships or attends school in a defined area; and membership of a group, such as a place of worship, a school, a labour union or a homeowners’ association.

Underneath those four routes sit three kinds of federal charter, and the charter decides which routes exist at a given institution. A single common bond charter serves one group related by occupation or association. A multiple common bond charter serves several such groups, each with its own distinct bond. A community charter serves people and organisations inside a well-defined local community, neighbourhood or rural district.

The practical reading is that the geographic route is the one that catches most people, and it is the one they least expect. A community-chartered credit union does not care where you work. It cares whether you live, work, worship or study inside a boundary, and boundaries are published.

Two cautions belong here. The first is that joining and qualifying are different events. Eligibility gets your application looked at; it does not underwrite you, and a credit union will decline a loan to a member as readily as a bank will decline a customer. What our page on how a credit file becomes a price describes happens at a credit union too.

The second is an absence, and we would rather name it than paper over it. NCUA publishes how many credit unions there are and how many members they have. It does not publish any figure for how many Americans are eligible to join how many institutions, and we have not found one anywhere we would defend. Anyone telling you that some percentage of the country qualifies for a credit union has produced that figure rather than retrieved it. The only reliable answer is the one you get by checking the two or three nearest to you against their own published field of membership.

Two charters, one word on the sign

Of the 4,250 federally insured credit unions counted at the end of March 2026, 2,672 held federal charters and 1,578 were state-chartered institutions that carry federal share insurance. Both are credit unions. Both are member-owned and not-for-profit. Both appear in the NCUA rate comparison. And the two of them are governed by materially different lending rules, which almost no consumer coverage mentions.

The federal lending rule is 12 CFR 701.21. Its own scope paragraph says the section generally applies to federal credit unions only, and that certain provisions reach federally insured, state-chartered credit unions as specified at 12 CFR 741.203. That second section is a short list, and the interest ceiling is not on it.

Here is what follows which charter, as the rules stood when they were read on 7 September 2026.

Nothing in that table says a state-chartered credit union will charge you more or treat you worse. State law imposes its own limits, and a state charter is not an absence of rules. What the table says is narrower and more useful: two of the protections most often attributed to “credit unions” in general are attributes of one charter in particular, so if either matters to you, the question to ask is which charter the institution holds — and the answer is usually in its name or on its own disclosures.

The ceiling a federal credit union cannot lend above

This is the most consequential thing on the page for anyone whose credit file is going to be the problem, and it is almost never mentioned in coverage of credit union car loans.

Except when the Board establishes a higher maximum, a federal credit union may not extend credit to members at rates exceeding 15% per year on the unpaid balance, inclusive of all finance charges. That is the statutory position. At least every 18 months the NCUA Board decides whether to raise it, and it may do so only after consulting congressional committees, the Treasury and the other federal financial regulators, and only if it finds that money market rates have risen over the preceding six months and that prevailing rates threaten the safety and soundness of individual credit unions.

The Board has kept a temporary ceiling of 18% in place. Its letter to federal credit unions in February 2026 records that the previously approved ceiling was due to expire on 10 March 2026 and that the extension runs through 10 September 2027. The same letter preserves a separate, higher allowance for payday alternative loans, on which up to 28% may be charged under the conditions NCUA’s regulations specify.

Two things follow, and they point in opposite directions.

The first is that a federal credit union has a hard ceiling on the price of the money, and the rest of the market does not. A subprime borrower quoted a rate at a finance company or through a dealership is not protected by anything comparable. Whatever a federal credit union offers, the all-in figure cannot exceed the ceiling in force, and finance charges count towards it rather than sitting outside it. At the difficult end of the used-car market, where the price of the money is the whole problem, a hard statutory limit on that price is a materially different product from one without any.

The second is that a ceiling is not a rate. Being told the maximum is 18% tells you nothing about what you will be offered, and it caps the price of the money only — not the price of the car, not the fees a dealer adds, not the products sold alongside the loan. It is a backstop against the worst outcome, not evidence about the likely one.

The prepayment rule sits alongside it and is easier to overlook. Under the same section, a member of a federal credit union may repay a loan, or an outstanding balance on a line of credit, before maturity, in whole or in part, on any business day, without penalty. On a five- or six-year car loan that is worth real money to anyone who expects to pay it down early or refinance out of it, and it is a rule rather than a courtesy.

Both of those carry dates. The ceiling expires on a stated day and is reviewed on a fixed cycle, so anything you read about it — including this page — has a shelf life. Check the current letter rather than trusting a figure with no date attached to it.

What 2.20 points is actually worth

A spread expressed in percentage points is not yet money, and the step from one to the other is where most rate comparisons quietly stop. The published gap between the two channels on a 48-month used-car loan was 2.20 points on that December Friday. What that is worth depends entirely on how much you are borrowing and for how long, which are facts about you rather than facts about the market.

The calculator below takes two quotes — the amount financed, and each offer’s APR and term — and returns what each one costs in total rather than per month, which is the comparison the paperwork does not make for you.

What it puts in front of you is the total interest, and the total is the only quantity in the transaction that gets worse as the term stretches while the monthly figure gets better. Two offers can sit two points apart and be closer in total cost than they look, if the cheaper rate comes attached to a longer schedule; two offers a fraction of a point apart on the same term are exactly as far apart as they appear. The point of running your own numbers is that neither outcome is predictable from the percentages alone.

One caution before you treat any gap as yours. The published averages are national averages of posted rates for a standard product. Your quote is priced against your file, the car, the term and how much is being advanced against what the lender believes the car is worth — and none of those is in the table. A spread in a national average is a reason to obtain a second quote, not a discount you can count on.

The gap is not a constant, and it has been close to nothing

Here is the finding that a single quarter conceals, and the reason this page reads the same row at seven year-ends rather than one.

Take the 48-month used-car line from each December comparison. On 27 December 2019, credit unions averaged 3.62% and banks 5.56%, a gap of 1.94 points. On 25 December 2020, 3.16% against 5.16%, a gap of 2.00. On 31 December 2021, 2.85% against 5.06%, a gap of 2.21 — the widest of the seven.

Then the price of money moved, and the gap did something the folklore does not anticipate. On 30 December 2022, credit unions averaged 4.79% and banks 5.86%: a gap of 1.07 points, roughly half what it had been a year earlier. On 29 December 2023 it narrowed again, to 6.45% against 7.21% — a gap of 0.76 points, the narrowest of the seven readings and about a third of the widest.

Then it reopened. On 27 December 2024, 6.11% against 7.84%, a gap of 1.73. On 26 December 2025, 5.53% against 7.73%, a gap of 2.20 — back within a hundredth of a point of the 2021 maximum.

The shape is unmistakable in the numbers even if the cause is not settled by them. Both columns climbed steeply into 2023, and the credit union column climbed faster — from 2.85% to 6.45% while the bank column went from 5.06% to 7.21% — which is what closed the gap. Afterwards the two came apart again: the credit union average fell back to 5.53%, while the bank average carried on up to 7.84% before easing only to 7.73%. That is the pattern you would expect if credit union pricing lags the market in both directions rather than tracking it — a reasonable reading of an institution whose own regulator describes its lending as one member’s savings becoming another member’s loan. But it is a reading, and the table does not prove it.

What the table does prove is the practical point. A used-car buyer at the end of 2023 who assumed a credit union would save them two points was working from a rule of thumb that was, that year, wrong by more than a point. The direction of the advantage held across all seven readings. The size of it did not hold at all.

Two federal publications, one market, two answers

There is a second reason to distrust any single quoted average, and it appears the moment you set the two federal sources against each other.

NCUA’s comparison puts the bank average for a 48-month new-car loan at 7.33% for the last Friday of December 2025. The Federal Reserve publishes its own commercial-bank series for exactly that product, and the readings either side of that date were 7.62% in November 2025 and 7.37% in February 2026. So NCUA’s figure for banks sits 0.29 points below the Federal Reserve’s nearest earlier reading of the same product, and 0.04 below its nearest later one. Same product, same class of institution, weeks apart, two federal publications, three different numbers.

Neither is wrong. They are different measurements. One is a vendor survey of rates institutions had posted; the other is the Federal Reserve’s own collection from commercial banks. The disagreement is the useful part, because it puts a floor under how precise anybody can honestly be: if two federal bodies describing the same product differ by roughly a quarter of a point, a page telling you the average used-car rate to two decimal places is claiming a precision the record does not support.

One more disagreement between them is worth stating, because it looks like a contradiction and is not. In the Federal Reserve’s most recent reading, for May 2026, the advertised rate falls as the term lengthens: the 48-month figure sits above the 60-month, which sits above the 72-month. The NCUA table, on its December date, does the opposite in both of its columns: used cars cost 5.41% over 36 months and 5.53% over 48 at credit unions, and 7.69% against 7.73% at banks; new cars cost 5.32% over 48 months and 5.44% over 60 at credit unions, 7.33% against 7.41% at banks. Longer is dearer, by 0.12 points at credit unions on both pairs and by 0.04 and 0.08 at banks.

The explanation is the one our page on used-car financing rates gives for the Federal Reserve’s ordering in the first place: those series average loans actually written, and a six-year loan is written against a different car for a different borrower than a four-year loan, so the populations differ. Hold the product constant — which is what a posted-rate table does — and the ordinary term premium reappears. Two publications, two orderings, one underlying fact about how lending is priced.

The same table settles something else the Federal Reserve series cannot, because it has no used-car line at all. In NCUA’s bank column on that date, a 48-month used-car loan cost 0.40 points more than a 48-month new-car loan. In the credit union column, 0.21 points more. Used paper prices above new in both channels, from a source that prints both, and the premium at banks was nearly twice the premium at credit unions.

A credit union loan arranged at the dealership is still a dealer-arranged loan

The averages above describe direct lending: a member goes to the institution, applies, and is quoted. That is not the only way a credit union’s money reaches a car buyer, and the difference is invisible on the contract.

Credit unions also buy paper originated at dealerships, and NCUA regulates the arrangement rather than forbidding it. Under 12 CFR 701.21(h), a federally insured credit union may not acquire vehicle loans serviced by a third-party servicer beyond 50% of its net worth during the first thirty months of that relationship, or 100% afterwards, without a regional director’s waiver. That rule exists because indirect vehicle lending is a real and substantial part of the business, not a fringe of it.

For a borrower the consequence is straightforward and easy to miss. If the finance office at a dealership tells you the loan is with a credit union, the loan may well be with a credit union — and the rate you are being shown was still assembled at that desk. Everything our page on used-car financing rates says about the gap between a lender’s buy rate and the rate presented to the customer applies to that transaction, because the arrangement is the same arrangement. The institution behind the paper does not change who added the margin.

Which means the published comparison is not evidence about the offer in front of you unless the offer came to you directly. Getting a direct quote, from the institution rather than through the sale, is what converts the national average from a statistic into something you can hold your dealer-arranged offer against — and our page on car loan pre-approval covers what that application involves and how long an offer lasts.

What this comparison cannot tell you

A page resting this heavily on one federal table owes an account of where it stops.

They are averages, with no dispersion attached. The table gives the middle and says nothing about the width. The distance between the best and worst rate written at credit unions in the same week is not in these numbers, and that distance is exactly what a borrower most wants to know.

They are posted rates, not written rates. The figures represent what active institutions had reported for one Friday. Nothing here says how much lending happened at those rates, or to whom.

It is a vendor survey inside a federal publication. The Call Report figures cited earlier — the number of institutions, the size of the used-car book — are a regulatory filing made by every federally insured credit union, and they are a census. The rate comparison is not, and the two should not be given equal weight because they arrived on the same website.

It lags, sometimes badly. On 7 September 2026 the most recent quarter NCUA had published was 2025 Q4, describing rates posted more than eight months earlier, while the Federal Reserve’s equivalent series already carried a May 2026 reading. The dating footnotes on the NCUA pages also contain at least one obvious slip — the December 2025 comparison records an extraction date in the January before its own reference Friday. Read the “For [date]” line at the top of the page, which is the reference date, rather than the extraction line beneath the table.

A national average is not an institution. There are 4,250 federally insured credit unions. Any one of them can price above or below the average for reasons that have nothing to do with you, and the only way to find out is to ask two or three of them.

It says nothing about the car. A rate is the price of the money. The collateral is underwritten separately, and a title brand, an odometer discrepancy or an unrepaired open safety recall will move a lender further and faster than any argument about a fraction of a point. That is true in this channel exactly as it is in every other.

Working the credit union channel

What follows from all of the above is a short and dull sequence, which is usually the sign that it is right.

One last note for a specific case. If the car you are buying belongs to a private seller rather than a dealer, the choice of channel narrows sharply and credit unions become disproportionately important — they hold much of that market and are used to the payoff and title mechanics it requires. Our page on private-party auto loans covers how that transaction is put together.

Common questions

Are credit union used car loan rates actually lower than bank rates?

On the federal comparison, yes, on every reading we checked. For rates posted on 26 December 2025, the 48-month used-car average was 5.53% at credit unions against 7.73% at banks, and the 36-month average 5.41% against 7.69%. The direction held at all seven year-ends we read, from 2019 to 2025. The size of the gap did not: it ranged from 0.76 points to 2.21 across those seven readings. These are national averages of posted rates on a stated date, not quotes.

What is the average credit union rate on a used car right now?

Nobody can tell you what it is right now, and the honest version of the answer is the useful one. NCUA publishes the comparison quarterly; when this page was checked on 7 September 2026 the most recent published quarter was 2025 Q4, giving 5.53% at 48 months. Anything more current than that is not from this source. Check the NCUA page itself for a newer quarter before relying on the figure.

Why are credit union rates lower?

Because of where the surplus goes. A credit union is owned by its members and is not-for-profit, so profits come back to members rather than to outside shareholders. NCUA names three forms that can take: reduced fees, higher savings rates and lower loan rates. Note that this is a choice among three, which is part of why the gap moves — and why 21 of the 23 products NCUA compares favoured credit unions on that date while two, both deposit accounts, did not.

Do I have to join a credit union before I can get a car loan?

Yes — lending is to members, which is what membership means here. Eligibility runs through a common bond: your employer, your family, a group you belong to, or the area you live, work, worship or study in. Many institutions let you join at the point of application. Being eligible is not the same as being approved; membership gets your application read, and the underwriting that follows is the ordinary kind.

Is there a cap on what a credit union can charge me?

At a federal credit union, yes. The statutory ceiling is 15% a year on the unpaid balance including all finance charges, and the NCUA Board may set a higher temporary rate, reviewed at least every 18 months. The Board has 18% in force, and its letter to federal credit unions in February 2026 records the extension running through 10 September 2027. A separate allowance permits up to 28% on payday alternative loans. State-chartered credit unions are not covered by that rule — 12 CFR 741.203 does not extend it — and are governed by their own state’s law. A ceiling is a backstop, not a forecast of your rate.

Can a credit union charge me a penalty for paying the loan off early?

Not at a federal credit union. Under 12 CFR 701.21(c)(6), a member may repay a loan or an outstanding line-of-credit balance before maturity, in whole or in part, on any business day, without penalty. That paragraph is not among the ones 741.203 extends to federally insured state-chartered credit unions, so on a state charter it is a question to ask rather than an assumption to make.

Is my money safe at a small credit union I have never heard of?

Deposits at federally insured credit unions are covered by the National Credit Union Share Insurance Fund, which is backed by the full faith and credit of the United States, up to $250,000 per share owner, per insured credit union, for each account ownership category. That covers all federal credit unions and the vast majority of state-chartered ones. It protects money you deposit; it has no bearing on a loan.

The dealer says my loan is with a credit union. Is that the same as going direct?

No. Credit unions buy vehicle paper originated at dealerships — NCUA caps how much of it any one third-party servicer can hold, at 50% of net worth for the first thirty months of the relationship and 100% after — so a credit union’s name on the contract does not mean the rate was set by the credit union alone. A quote arranged at a dealership is a dealer-arranged quote whoever funds it. The published averages describe direct lending, and a direct application is the only thing that produces a number you can compare them to.

How much older is a used car allowed to be for a credit union loan?

Neither of the federal sources on this page answers that, and we are not going to invent a number. Age and mileage limits are set institution by institution, vary widely, and are usually published in the lender’s own loan policy or disclosed on application. What can be said from the data is that used-car lending is not a marginal activity for this channel: $321.3 billion of it was on the books at the end of March 2026, 67.0% of all the vehicle lending these institutions had done.

Does a credit union quote still include a dealer markup if I refinance later?

A refinance obtained directly from a lender has no dealer in it by construction, which is the structural difference rather than a promise about the rate. Whether refinancing is worth doing depends on the balance, the remaining term, and what the car is worth against what you owe — none of which this page can see.