How Many Shelby Cobra Replica, Kit, Tribute, and Continuation Cars Change Hands Every Year?

This article answers one question: how many Shelby Cobra replica, kit, tribute, and continuation cars are sold in a year? Nobody publishes that number, so I built it from the ground up.

Roughly 1,300 of these cars sell in North America each year, and roughly 1,750 are offered for sale. The defensible range on the first figure is about 850 to 2,000.

That estimate is built from the population of cars, not from sales records, for the simple reason that no complete sales record exists. I then checked it against the one slice of this market that is publicly documented, which is what the two largest auction houses report. The two approaches agree on the order of magnitude and disagree by about 40 percent in the middle, which is roughly what honest triangulation looks like.

What follows is the estimate, the check, and the specific places each could be wrong. I have shown the math so you can disagree with the assumptions rather than the conclusion.

First, a boundary problem worth naming

The obvious way to ask this question is “how many sell in the U.S.” That phrasing does not survive contact with the data, so I want to be precise about what I am counting before I count it.

The estimate rests on a population figure, and every serious estimate of this population is a North American one rather than a U.S. one. Mixing a U.S. numerator with a North American denominator would produce a number that looks precise and means nothing. So the unit is North America.

1. A sale counts if the car was offered for sale in the United States or Canada.
2. It counts where the car was offered, not where the buyer lives or where the car ended up. Exports still count. Imports sold here still count.
3. The U.S. is the overwhelming majority of it. In the auction records I use later as a check, better than 98 percent of the cars were located in the United States and about 1 percent in Canada. Everything else was a rounding error.

There is a market outside North America as well. When I catalogued who has built these cars, more than a third of the companies were outside North America, across the United Kingdom, South Africa, Australia, and a dozen other countries. This article does not attempt to size that market. I have no basis for it, and I would rather say so than extrapolate.

The primary method: start from the cars, not the sales

If you cannot observe transactions, count the population and estimate how often it turns over. That is the standard approach for any asset class where most trades happen privately, and it has one large advantage here. It does not depend on any single company’s records.

It has one large disadvantage too, which I will not hide: the population figure is the softest number in this article. Here is exactly how soft, and why.

What the builders actually publish

Almost nothing. I went looking for cumulative production figures from every significant builder and found two.

1. Factory Five Racing reports more than 16,000 kits sold. That figure covers all eleven of its products rather than Cobras alone, so it cannot stand as a Cobra count. It is the clearest evidence of volume leadership and a good example of why a precise fleet number stays out of reach.
2. Superformance reports more than 3,000 MkIII Cobras delivered, against more than 6,500 cars across all its models. This is the only clean, Cobra-specific cumulative figure I could find from any builder.

Every other name on a roster of more than 130 companies publishes no cumulative total at all. Several widely repeated figures turn out to have no source when you chase them, and I have not confirmed production numbers for ERA, Kirkham, Backdraft, or Hi-Tech from anything I would be willing to cite. A handful of defunct builders left better records than the live ones, because somebody wrote a history when the company closed. Aurora in Canada built roughly 170 cars. Autokraft in the United Kingdom built somewhere between 450 and 480.

The honest summary is that this market cannot be counted from published builder figures, and anyone who hands you a precise fleet number is guessing.

What can be said about the population

Working from the two published figures, the shape of the roster, and the share of that roster that is North American, I put the North American fleet at 25,000 to 40,000 cars, with 30,000 as the middle case. That is a judgment estimate and I am labeling it as one. It sits below the 60,000 to 80,000 usually cited worldwide, which is the sanity check it has to pass.

There is one independent signal worth putting next to it, and it is more useful as a warning than as a confirmation. FFCars, the largest Factory Five forum, reports about 48,800 registered members, and Club Cobra has run to tens of thousands. Those numbers are larger than my entire estimated fleet. That is not evidence the fleet is bigger. It is only evidence that forum registrations count lurkers, dreamers, former owners, and dead accounts, and that no online population can be read as an owner count. I include it because it is the kind of number people may know, and it does not work.

Turnover

Collector vehicles change hands at something like 3 to 6 percent a year. These cars sit at the driver end of that world rather than the museum end, so I have used 4 to 6 percent, which leans slightly toward the active side of the band. This is an applied assumption, not a published figure for this category.

The primary estimate

On its own, that method says about 1,500 sales a year. Now the question is whether anything independent agrees.

Why most of these sales are invisible

Before the check, it is worth establishing what a check can and cannot see, because the answer determines how much weight the public record can carry.

Most of these cars trade in private. They change hands owner to owner. They move through club rosters, through marque forums, through the Facebook groups where a car gets posted on a Tuesday and is gone by Friday, and through specialist dealers. Many of them never leave the circle of people who have known each other for twenty years. Even my own car was an owner-to-owner sale, and I found it by posting on the ERA Facebook group.

That is not a guess. Hagerty’s John Wiley reported in March 2023 on the January that had just closed, the biggest month for collector car auctions on record. Nearly 7,000 vehicles sold in person and nearly 2,000 more online. In the same span, nearly twice as many changed hands privately. That was the peak auction month of the year, so in an ordinary month the private share is larger still. Wiley’s more specific finding matters even more here. For vehicles under $100,000, he wrote, private transactions continue to rule the day. The median car in this category sits around $56,000, squarely inside that band.

So any public sales record is a minority slice by construction. That is the correct expectation to hold before looking at one.

The check: what the public record shows

The publicly documented portion of this market is auction results, and the two largest venues for these cars are Bring a Trailer and Mecum. Both publish their outcomes. I compiled every Shelby Cobra listing from each — 1,345 vehicles at Bring a Trailer from February 2015 through July 2026, and 619 at Mecum from 2014 through 2026 — then removed non-vehicle lots and the twelve genuine original Shelby and AC Cobras that appear across the two, which belong to a different market. That leaves 1,336 and 616 respectively in scope.

Over the four complete years from 2022 through 2025, the two houses together brought 1,139 of these cars to market and sold 845 of them. An average of 285 offered and 211 sold per year.

Source: author’s compilation of published auction results, 2022–2025. Excludes original Shelby and AC Cobras.

Two findings come out of that record, and one of them is worth more than the volume figures.

The first is that roughly one car in four does not sell. Every published count of this market I have seen counts cars that came to market, which is not the same as cars that sold. Count listings instead of sales and you overstate the market by about 35 percent. This is the distinction that makes the difference between the two headline numbers at the top of this article.

The second is that the two houses clear at nearly the same rate. Over four years one cleared 74.2 percent of the Cobras it listed and the other cleared 74.1 percent. Those are two different businesses. One is a seven-day-a-week online timed auction with a comment thread and an international audience. The other is a live block in a convention center with a ring man and a crowd. They converged inside a tenth of a percentage point across more than a thousand cars, which is a stronger result than either number alone and the best reason to think the rate reflects the market rather than either venue.

The rate also moves with the market.

Source: as above, both venues combined.

That is the pandemic collector-car boom and its unwind in a single column. In 2021 more than nine of every ten cars offered found a buyer, which is a market where sellers set reserves below what the room would pay. By 2023 it was back to seven in ten, and it has settled in the low seventies since. A 74 percent clearance rate means buyers and sellers disagree about value roughly a quarter of the time, which is a normal functioning market rather than a hot one or a broken one.

Turning the check into a comparable number

The 211 counted sales are a floor, not a market. To compare them against the population estimate they have to be stepped outward twice.

The first step covers what these two records miss: Barrett-Jackson, GAA, Hemmings, Cars and Bids, the Bonhams and RM tail, and specialist dealer inventory that never touches an auction block. The two largest venues run roughly 60 to 65 percent of tracked public volume in this category, which is what puts the uplift in that range.

The second step is the load-bearing assumption in the entire check. Auctions and dealers account for something like 15 to 30 percent of collector-vehicle units generally. I have set this market lower, at 18 to 35 percent with 25 percent as the middle case, for the reason Wiley identified. These cars sit under $100,000 where the private market rules, and they sit inside an unusually strong club and forum culture on top of that.

Do the two agree?

Best estimate: roughly 1,300 replica, kit, tribute, and continuation Cobra sales per year in North America, with a defensible range of about 850 to 2,000.

The two approaches share no inputs. One starts from a population and works inward. The other starts from published outcomes and works outward. They disagree by about 40 percent at the middle setting, with the public record running lower. They agree on the order of magnitude, which given how different the two paths are is the meaningful result.

The last row is a consistency check and it is the reason I trust the blend as far as I do. Divide the blended answer back into the installed base and the implied turnover lands between 3.4 and 5.0 percent, inside the collector-vehicle band the primary method assumed rather than contradicting it. If the blend had implied 12 percent turnover, one of the two methods would be wrong and I would not be publishing either.

The same market, counted as cars offered

A seller or a builder cares how many cars come to market, not only how many close. The clearance rate provides the conversion.

If the private market clears at the same 74 percent the two auction houses do, then about 1,750 of these cars are offered for sale across North America in a year, against roughly 1,300 that sell, with a range of about 1,150 to 2,700 offered.

Treat that as an illustration rather than a fourth measured figure, and note that it is the one number here that genuinely depends on the auction record. Everything above it survives without that data. This does not. Nobody can observe the private no-sale rate, because a car listed in a Facebook group in March and quietly relisted in July leaves no record that it failed the first time. There is a reasonable argument that private sales clear better than auctions, being negotiated rather than reserve-gated, and a reasonable argument that they clear worse, because sellers without a deadline hold out longer. I have applied the only rate I can measure and labeled the assumption.

What they bring

Price is not the subject of this article, but it sets the scale of the market and it takes two paragraphs.

Over 2022 through 2025, the median replica sold at auction for $55,500 at one house across 614 sales and $60,500 at the other across 167. Licensed Shelby CSX continuation cars run in a different bracket entirely, with medians of $170,000 and $143,000 across 60 and 34 sales respectively.

The direction has been up, and less dramatically than the hobby tends to repeat to itself. The same replica median sat at $43,375 across 2017 through 2019, which puts the gain at about 28 percent over six years. Two caveats belong with that. The early sample is thin, at 26 sales against 614, and the rise is not continuous. The median climbed every year through 2023, then eased to $56,500 in 2024 and $52,500 in 2025.

Where this is weakest

Five things could move these numbers, and I list them here rather than bury them.

1. The installed base is soft by about 40 percent, and it now carries the primary method. It is the single largest source of error in this article. Move it from 30,000 to 25,000 and the primary estimate drops by 500 sales a year. The clean fix is an insurer’s count of these cars in force. No insurer publishes one that I could find.
2. Turnover is applied, not measured. The 4-to-6-percent band is a general collector-vehicle assumption. Nobody has published a turnover figure for this category.
3. The private share in the check is an assumption too. The Hagerty reporting establishes that private sales dominate the collector market broadly and dominate below $100,000 specifically. It does not establish the exact ratio for this category. Move that input from 25 percent to 30 percent and the check falls by about 175 sales a year.
4. The auction records count events, not cars. Results pages do not show chassis numbers, so a car sold in 2022 and again in 2025 appears twice with no reliable way to link the two. For a question about transactions per year that is correct behavior. It does mean the check cannot tell you how many distinct cars exist.
5. The auction compilation is a search result, not a census. It reflects lots listed under “Shelby Cobra.” Cars catalogued only under a builder’s name would be missed, which biases the counted floor low rather than high.

One more limit belongs here. Both compilations were built by hand from each house’s own published results, and transcribed rather than downloaded, because neither offers the data in any other form. I have spot-checked rather than verified all 1,952 rows. Individual cells may be wrong. The aggregates are sound.

What the numbers actually mean

Take the estimate at face value and the picture is a market of roughly 30,000 cars in North America. About 1,750 of them are offered in a year and about 1,300 sell, at a median in the mid-fifty-thousands. Roughly five in six of those sales happen somewhere no one can observe.

Three implications follow, and the last is the one I care about.

The first is that this is a modest market by collector-car standards. About 1,300 transactions a year is a real market and a small one.

The second is for anyone selling one. A quarter of the cars brought to auction go home unsold, and the records do not say why any individual car failed to clear. What they do establish is that the rate is steady at roughly one in four, across both venues and four years. That is worth knowing before you pay to put a car in front of a room.

The third is structural. The majority of these cars trade in exactly the conditions where a documented record is worth the most and is least likely to exist. At an auction house, a professional writes the description, photographs the car, and stakes a reputation on the listing. In a private sale between two strangers who met in a Facebook group, the entire evidentiary basis for a $60,000 transaction is what the seller says and what he can produce from a folder. I have argued before that a replica’s record has to be authored by its caretakers because no factory or institution authored one for it. The transaction estimate says where that authoring pays off. It is not the 211 sales a year with a catalogue entry behind them. It is the roughly 1,100 without one.

A car is only worth what can be proven about it.

If a figure here is wrong, I want to know. The assumptions are the part most worth arguing with, and there are four of them doing real work: the fleet size, the turnover rate, the private-market share, and the clearance rate applied to private sales. If any of those looks wrong to you, tell me why, and tell me what you would put in its place.

Method and sources

• Installed base of 25,000 to 40,000 in North America: the author’s judgment estimate, built from published builder figures, the composition of a roster of more than 130 companies, and the commonly cited worldwide figure of 60,000 to 80,000. Not a sourced count, and labeled throughout as an estimate.
• Builder production figures: Factory Five Racing (16,000-plus kits, all products, self-reported) and Superformance (3,000-plus MkIII Cobras and 6,500-plus cars across all models, company-stated). No other builder publishes a cumulative total. Aurora (approximately 170) and Autokraft (450 to 480) are historical figures for defunct companies.
• Forum populations: FFCars (approximately 48,800 registered members) and Club Cobra, cited as an upper bound and explicitly not as an owner count.
• Annual turnover of 4 to 6 percent: general collector-vehicle assumption, applied by the author. Not a published figure for this category.
• Private-market dominance: John Wiley, “The state of 2023’s private collector-car market,” Hagerty Media, 3 March 2023. https://www.hagerty.com/media/market-trends/the-state-of-2023s-private-collector-car-market/
• Auction check: 1,345 Shelby Cobra vehicles listed at Bring a Trailer February 2015 through July 2026, and 619 at Mecum from events 2014 through 2026, compiled by the author from each house’s published results in July 2026. Non-vehicle lots and twelve original Shelby and AC Cobras removed, leaving 1,336 and 616 in scope. Volume, clearance, and price figures calculated over the four complete years 2022 through 2025.
• Blended estimate of about 1,300 sold per year: the author’s triangulation of the two methods above. The 1,750 offered figure applies the measured 74.2 percent auction clearance rate to the whole market.

Assumptions and limits

1. The primary estimate rests on a fleet figure that is soft by roughly 40 percent and a turnover rate that is applied rather than measured. These are the two inputs most likely to move the answer.
2. The auction check captures public sales only. Private, club, forum, and dealer transactions are estimated rather than counted.
3. The cars-offered figure applies the measured auction clearance rate to the private market, where the true rate is unobservable. It is an illustration, not a measurement.
4. Auction listings are counted as events. Without chassis numbers, a car sold twice in the period cannot be distinguished from two cars, so the check measures transactions rather than unique vehicles.
5. Builder attribution comes from listing text. Cars whose listings named no builder are uncategorized.
6. One house records the event rather than the date, so the year is exact and month-level timing is approximate.
7. Both auction compilations were transcribed by hand from published results pages. Spot-checked, not exhaustively verified row by row.
8. This article does not size the market outside North America.

A note on the sources. The auction results used as a check belong to Bring a Trailer and to Mecum, and both houses deserve the credit for publishing them at all. Nothing here reproduces their listings, photographs, or descriptions. What is published are counts, medians, and rates I calculated from publicly posted outcomes, which is the same thing any market analyst does with auction results. The compilations themselves are not published. If either house would like something here corrected or removed, write to me and I will handle it.

Disclosure. I own an ERA 427 S/C Shelby Cobra replica. I have no commercial relationship with Bring a Trailer, Mecum, Hagerty, or any manufacturer named here. This analysis follows the evidence and takes no vendor money.