How many Commodore 64s sold

Last Updated on August 23, 2026 by Dave Farquhar

The Commodore 64 is the best selling computer of all time. Depending on who you ask, it sold 17 million, 20 million, or 12 million units. And the generally accepted sales figure for the Commodore 128 is 4.5 million units. Who is correct? How many Commodore 64s and Commodore 128s sold?

The commonly repeated figures of 17 million Commodore 64s and 4.5 million Commodore 128s don’t line up with Commodore’s own annual reports and other internal sales documents. The primary sources indicate Commodore sold about 12.3 million Commodore 64s and 128s between 1982 and 1993.

Commodore’s primary sources say 12.3 million units

It took a number of years, but primary sources have surfaced. Commodore historian Dave McMurtrie made a YouTube video about Commodore sales figures. He took his numbers from internal documents that he received from former Commodore executives Don Greenbaum and Michael Tomczyk.

He also broke out the individual figures in a blog post, in tabular form.

Based on the numbers in those documents, he came up with a figure of 12,350,000 combined sales for the Commodore 64 and 128. Commodore never reported Commodore 128 sales figures separately. The 4.5 million figure that tends to get repeated a lot is impossible. McMurtrie suggested 2.5 million as a generous estimate. Truth be told, the real number of 128s sold was probably closer to 1.75 million. And that would leave the total number of C-64s sold at about 10.6 million.

McMurtrie did find an error in one of the annual reports, so it is possible these numbers are not spot on, but there is no way to get from the numbers in the annual reports to 17 million.

There is one year we don’t have information for. Commodore did not survive fiscal year 1994, so we don’t have any sales figures for its final year. However, sales for that year would have been a rounding error. After selling 650,000 units in fiscal 1992, sales tailed off to 200,000 units in fiscal 1993. Demand for 8-bit machines was fading in the 1993/94 time frame. Commodore survived approximately half a year, so if they had kept up the 1993 pace, the most they could have sold is 100,000 units. Given they were competing with inexpensive 386-based machines that year, managing to sell 100,000 units seems overly optimistic. We know they didn’t sell zero, because Commodore was throwing together whatever they could to try to stay afloat a little while longer. But whatever they sold was from whatever leftover parts they had on hand.

So somewhere between 12 and 12.5 million units would be about right, based on the numbers Commodore reported. And it’s still an astounding number of machines in a time when the majority of people wondered what they would do with a computer in their home.

Some highlights from the video: the best year, sales wise, was fiscal year 1984, with 2.5 million units. This probably won’t surprise many people. Fiscal years 1985 and 1986 followed with 2.0 million and 1.9 million, respectively. Sales tailed off to 1.1 million in fiscal year 1987, and Commodore ended the 1980s with 1 million each in fiscal 1988 and 1989. Sales in the ’90s were more modest, numbering in the hundreds of thousands per year.

More units sold in 1991 than in 1990. That seems odd, but after the Soviet Union fell, it became easier to sell the C-64 in Eastern Europe, and there was some pent-up demand for it. That caused a small surge in fiscal 1991 and helped prop up demand in fiscal 1992 before demand tailed off noticeably in fiscal 1993.

It’s also noteworthy that the 64 sold pretty well right out of the gate, at its introductory price of $595, selling half a million units in its first 6 months. 1982 was a good year for computer sales anyway. In 1982, Time observed computer sales doubled every year from 1980 to 1982.

Commodore’s fiscal year

One more thing about Commodore’s fiscal year. Its fiscal year ran from July 1 to June 30. So that’s why there were zero C-64 sales in fiscal 1982. When Commodore booked its first C-64 sales in August or September 1982, it was already fiscal 1983.

That’s also why Commodore didn’t list any C-128 sales for fiscal 1985. Commodore’s reporting lagged the calendar year by about six months.

This lag in the fiscal year also means we lost the 1994 sales, since Commodore didn’t survive long enough to issue an annual report for 1994. But it’s possible there were zero sales that year, or close to it. I say that because over the course of 1993, Commodore sold off a tremendous amount of inventory to liquidators, including most of the C-65 prototypes. This would have reduced the amount of C-64 inventory Commodore could have had on hand.

Commodore sometimes received C-64s as trade-ins, which it would then refurbish and resell. That activity would have continued into 1994, but those wouldn’t have counted as new sales.

Where the 17 million figure came from and why it stuck

There’s a reason that 17 million figure got repeated so much and eventually published in the Guinness Book of World Records. It’s also possible to figure out where some other numbers like 20 million and 21.5 million come from, and why they might sound like reasonable estimates.

Better information exists today. The reasons the sales figures were so elusive were because the data in Commodore’s annual shareholder reports was incomplete. And the data in the annual reports didn’t always match what Commodore’s marketing department was telling the press. To a degree, you expect that. The annual report covered fiscal years, where marketers could report calendar years or fiscal years depending on which number sounded better at the time. Marketers also didn’t have to account for returns or refurbished units.

Even Mark Brown, the managing editor of INFO magazine, was saying in the 2008-2010 time frame that he thought the figures magazines like his published in the 1980s were inflated. In the September-October 1988 issue of INFO, Loren Lovhaug wrote, “Commodore has sold 10 million 8-bit computers. There are now eight million C64s and two million C128s in the world.” The actual number of C-64s and 128s at that point was closer to 7.5 million and 1.5 million. Commodore probably had sold 10 million 8-bit computers by mid-1988, but only if you included the VIC-20.

Sometime around 1987, the Tandy 1000 overtook the Commodore 64 in annual sales to become the bestselling home computer in the United States. But its total sales figures never caught up, partly because of the huge number of units Commodore sold in 1983 and 1984. No single model of PC outsold the C-64, although PC sales combined, if you include compatibles, did eclipse the C-64.

But let’s look into why those numbers have so stubbornly stuck over the years.

Jack Tramiel himself

Jack Tramiel did not dispute the 17 million figure. Late in life he got that question a lot. And there was a good reason he had no reason to dispute the count of 17 million.

He left Commodore in 1984, so he wasn’t there to see its entire production run. He saw the beginning, and then he spent nearly a decade competing with it when he was running Atari. So he had an idea how many units were selling after he left. Of the three most obvious people to ask, he was by far the most honest. Nobody was going to get a straight answer out of Irving Gould or Mehdi Ali.

When asked, Tramiel repeated the generally accepted answer. Under his tenure at Commodore, the machine sold around 2 million units per year. Extrapolating that over a 10-year run, knowing that sales tailed off toward the end, 17 million probably sounded about right. Especially given by 2007, it had been 15 years since he had needed to know that information.

So while Jack Tramiel was confirming incorrect numbers, I don’t think he was being dishonest. The number would have seemed reasonable to him.

The infamous Usenet post

Dave McMurtrie traces the commonly accepted figure to a Usenet post from 2002. Where did the numbers in that post come from? Not from Commodore’s 1993 annual report as claimed. But I can walk through the steps to get to those numbers.

Let’s start with the C-128. Add up total sales from the annual reports during the C-128’s lifetime, and Commodore sold a total of 4.5-5 million C-64 and 128s in those years. Misread or misunderstand that number as C-128 sales, report the lower number to play it safe, and there’s your 4.5 million.

Where the C-64 number came from is trickier. But a series of rounding errors and too-generous estimates can get you to 17 million. Let’s step through that process.

In its best year, fiscal 1984, Commodore sold 2.5 million C-64s. Round up to 3 million. Running total: 3 million.

In its second best year, Commodore sold 2 million C-64s. Running total: 5 million.

The first fiscal year the C-128 was on the market, Commodore the 64 and 128 combined to sell 1.9 million units. Round up to 2 million and don’t worry about double-counting 128 sales. Running total: 7 million.

Confuse fiscal years and calendar years and remember that magazines like Compute!’s Gazette reported a couple of times in the late 1980s that Commodore had “sold about a million C-64s” the prior year. Consider that the C-64 was on the market from 1982 to 1994 and that you’ve only accounted for 3 of those 12 years. Multiply 1 million by 9 to “conservatively” estimate those missing years. Running total: 16 million.

Add another million since that was a conservative estimate in your mind. Final total: 17 million.

There are two problems with that number. You’re double-counting during the C-128 era. And then you’re overestimating sales in 1982, 1992, 1993, and 1994. Demand was high in 1982 but Commodore was ramping up production. While demand remained surprisingly high in 1992 and beyond, it was no longer a million units a year.

But the number sounds reasonable to anyone who wants to believe the C-64 sold well for a long time. So that’s why numbers like 17, 20, and 21.5 million have stuck over the years. It’s a series of double counting and rounding errors.

Dave Farquhar

David Farquhar is a computer security professional, entrepreneur, and author. He has written professionally about computers since 1991, so he was writing about retro computers when they were still new. He has been working in IT professionally since 1994 and has specialized in vulnerability management since 2013. He holds Security+ and CISSP certifications. Today he blogs five times a week, mostly about retro computers and retro gaming covering the time period from 1975 to 2000.