Women make up 20.3 percent of software developers in the United States, according to the Bureau of Labor Statistics Current Population Survey (2025 annual averages, published February 2026). Widen the lens to all computer and mathematical occupations and the figure rises to 27.5 percent.
So the short answer is roughly one in five, and the honest answer is a range of about 20 to 27 percent depending on which population you count.
That is not the number most people have heard. The number most people have heard is 10 percent, and it has been repeated so often that it has hardened into a kind of folk statistic. It is wrong. Or rather, it is measuring something else, and working out what that something else is turns out to be the most interesting part of this whole question.
One disclosure, because it is relevant. When we first published this article in 2023, we put that 10 percent figure in the headline. We were quoting the same sources everybody else was quoting. We have rewritten it from the ground up. Every number below links to the organization that produced it.Here is the data that settles most of the argument, ordered from the highest share of women to the lowest. All figures from BLS Current Population Survey Table 11, 2025 annual averages.
Web and digital interface designers — 57.0% women, 69,000 employed
Operations research analysts — 53.9% women, 178,000 employed
Software QA analysts and testers — 50.6% women, 72,000 employed
Computer systems analysts — 44.0% women, 569,000 employed
Web developers — 36.8% women, 66,000 employed
Database administrators and architects — 36.3% women, 148,000 employed
Computer support specialists — 29.1% women, 677,000 employed
Computer and information systems managers — 25.1% women, 646,000 employed
All computer and mathematical occupations — 27.5% women, 6,711,000 employed
Read that top to bottom before you quote any single number out of it.
Software quality assurance is 50.6 percent women. Software development, in the same occupational family, working on the same products, frequently in the same building, is 20.3 percent. Computer network architecture is 8.0 percent.
That is a spread of 49 percentage points inside a category the public hears about as a single number called "tech."
A methodological note, because it matters and nobody else will mention it
BLS states that its 2025 annual estimates are 11-month averages that exclude October, because data for that month were not collected during the federal government shutdown. The 2025 figures are therefore not strictly comparable with annual averages from other years.
The gaps described above are far too large to be an artefact of one missing month. But the caveat belongs in the article, and it belongs on any chart built from this table.
Why the sources disagree
Three different things are being measured, and they are routinely reported as though they were one thing.
The job title
Occupational data counts people who hold a specific job. Software developer, computer programmer and software QA analyst are three separate occupation codes in the BLS classification. They pay differently, they skew differently, and adding them together produces a different answer than counting any one of them. The table above is what that looks like in practice.
The industry
"Tech workforce" is a much wider net than "software engineer." It takes in product managers, designers, analysts, IT support and sales engineers. Women are better represented in several of those roles. That is why the all-occupations figure of 27.5 percent sits above the developer-specific figure of 20.3 percent, and why CompTIA and Deloitte both land in the mid-twenties when they report on "tech."
Both numbers are correct. They answer different questions, and they get quoted interchangeably by people who have not noticed.
Who answered the survey
This is where the 10 percent comes from, and it is the part nobody explains.
The Stack Overflow Developer Survey is the most widely quoted source in the industry, and it has consistently reported a female share far below the labour-force data. It is also, by its own account, not a representative sample. Stack Overflow's published methodology states that respondents were recruited primarily through channels Stack Overflow owns, which means highly engaged Stack Overflow users were more likely to see the prompt and take the survey.
That is not a criticism of the survey. It is the survey being straight with you about what it is.
But it means the number measures participation in one developer community. It does not measure employment. If women are underrepresented among the people who write software, and then underrepresented again among the subset of those people who answer Stack Overflow surveys, you get a figure that compounds two separate gaps and reports them as one.
Where the "10 percent" figure came from
Survey data being read as employment data. That is the whole answer.
It is worth sitting with, because the figure has done real damage on its way around the internet. A sixteen-year-old deciding whether to study computer science, reading that she would be one engineer in ten, is being told something considerably bleaker than the truth. The truth is one in five. One in five is a different kind of number to walk into a room with.
Bleak numbers travel further than accurate ones. They make a better headline, and nobody fact-checks a statistic that confirms what they already suspected. We know. We published one.
The URL of this page still says 10 percent. We left it there. A broken link helps nobody, and quietly deleting the record seemed worse than leaving it visible.
Fewer computer science degrees
In 1984, women earned 37.1 percent of the computer science bachelor's degrees awarded in the United States, according to the National Center for Education Statistics. That was the peak. It has never been matched.
What followed was a collapse that ran for a quarter of a century. By 2011 the share had fallen to 17.6 percent, the lowest since the mid-1970s. It has recovered slowly since, to 22.6 percent in the 2021–22 academic year, the most recent published.
Read that sequence again, because it dismantles most of the explanations people reach for. The share of women in computer science did not fail to grow. It grew, and then it fell, during precisely the decades when women were entering law, medicine and the physical sciences in rising numbers. Whatever happened to computing was specific to computing.
The most credible account is the one NPR's Planet Money reported in 2014: the arrival of the home computer. Early personal computers were marketed almost entirely to boys, as toys. Researcher Jane Margolis, interviewing computer science students at Carnegie Mellon, found that families were far more likely to buy a computer for a son than for a daughter, even when the daughter was the one who was interested. A decade later, those boys arrived in the first-year CS lecture having already spent five years at a keyboard. The girls arrived having not. And the professors, reasonably enough, taught to the room in front of them, which already knew how to program.
It is a hypothesis, not a proof. But it fits the timing better than anything else on offer, and the timing is the thing that needs explaining.
The broken rung
Culture is difficult to measure, which is why people who would rather not discuss it tend to point out that culture is difficult to measure. Several things around it are measurable.
The clearest is what LeanIn and McKinsey call the broken rung. It is not the glass ceiling and it is not the boardroom. It is the first promotion, from individual contributor to manager. In the 2025 Women in the Workplace report, for every 100 men promoted into that first management job, only 93 women were. For women of color, 74. Everything downstream compounds from there. Men outnumber women at manager level from the start, so the pool of candidates for the level above is already skewed, and it stays skewed all the way up. Women hold 29 percent of C-suite roles, a figure that has not moved since 2024.
This is why fixing the top of the pipeline has never worked. The break is at the bottom.
The BLS data shows the same shape inside tech. Computer and information systems managers, the closest thing the government tracks to an engineering leadership role, are 25.1 percent women. Higher than software developers at 20.3 percent, which looks encouraging for about four seconds, until you notice that those managers are drawn from the whole of IT, not just from engineering.
Is it getting better?
Three trends, and they are not moving in the same direction. Anyone who answers this question with one word is flattening something.
Employment: yes, slowly. The share of women in computing occupations has climbed over the last decade. It has not climbed fast, and the tech layoffs of the last few years took some of it back.
The degree pipeline: recovering, from a long way down. Up from 17.6 percent to 22.6 percent. Still fourteen and a half points below where it stood in 1984, which is to say that a woman graduating in computer science today is entering a field that is less female than the one her mother could have graduated into.
Leadership: stalled. 29 percent of the C-suite, unchanged year on year, and the broken rung is still broken. This is the one that matters most, because it is the lever that moves the other two.
What we see, from inside a women-owned engineering company
I have been running a software company for fourteen years and I have hired a lot of engineers. Here is the part the statistics do not capture.
The pipeline argument, the one that says there are simply not enough women applying, is true. It is also an excuse. It is true because the degree data is what the degree data is. It is an excuse because most companies never test it. They post a role, receive a pool that looks like every other pool, hire from it, and conclude that the market gave them no choice. Few of them ask why their pool looks like that, or where they advertised, or what their job description signals, or who sat on the interview panel.
Look again at the BLS table. Software QA is half women. Software development is a fifth. These are adjacent jobs, requiring overlapping skills, frequently on the same team, often in the same standup. Whatever is happening between those two numbers is not a pipeline problem.
It is a hiring problem, a promotion problem and a culture problem, and those are all problems companies own.
JetRockets is women-owned, and Clutch ranks us amongst the top three women-owned software development companies in the United States. I did not set out to build a case study. I set out to build a Rails team that ships. What I will say, after fourteen years of it, is that the composition of the room changes what gets built in it, and I have never once had to trade quality to get there.
If your engineering team has no women on it, that is a fact about your hiring. It is not a fact about the labour market.
Conclusion
There is no single reason women are underrepresented in software engineering. There is a degree pipeline that collapsed in 1984 and never recovered, a promotion structure that stalls women at the first rung, a leadership layer that has not moved in years, and a definitional fog that lets everybody argue from a different number.
The number, for the record, is 20.3 percent. It is not 10.
Getting it right is not pedantry. Every one of those problems is easier to argue about in good faith when the people arguing agree on what is true.
Frequently Asked Questions
What percent of software engineers are female? Women make up 20.3 percent of software developers in the United States, according to the Bureau of Labor Statistics Current Population Survey (2025 annual averages). Across all computer and mathematical occupations the figure is 27.5 percent. The honest answer is a range of roughly 20 to 27 percent, depending on which population is being counted.
Why do different sources give different percentages? Because they measure different things. Occupational data counts people who hold a specific job title, and software developer, computer programmer and QA analyst are three separate BLS occupation codes with markedly different gender splits. "Tech workforce" is wider still and includes non-engineering roles where women are better represented. And developer community surveys, such as the Stack Overflow Developer Survey, measure who chose to answer a survey rather than who holds the job. Stack Overflow's own methodology notes that respondents were recruited through channels it owns, which skews the sample toward its most engaged users. That is where the lowest figures come from.
Which tech jobs have the most women? Web and digital interface designers (57.0 percent), operations research analysts (53.9 percent) and software quality assurance analysts and testers (50.6 percent) are all majority or near-majority women. At the other end, computer network architects are 8.0 percent women and information security analysts are 15.9 percent. Software developers sit at 20.3 percent. All figures from BLS Current Population Survey Table 11, 2025 annual averages.
What percentage of computer science majors are women? 22.6 percent of computer science bachelor's degrees awarded in the United States go to women, as of the 2021–22 academic year. That is up from a low of 17.6 percent in 2011, but still well below the peak of 37.1 percent, which was reached in 1984 and has never been matched since.
Is the percentage of women in software engineering going up? It depends which number you mean. Employment is rising slowly. The degree pipeline is recovering from its 2011 low but remains fourteen and a half points below the 1984 peak. Leadership is stalled: women hold 29 percent of C-suite roles, unchanged since 2024, and only 93 women are promoted to first-line manager for every 100 men.
What is the most female-dominated engineering field? Among engineering occupations where BLS publishes a gender breakdown, chemical engineering has the highest share of women at 24.9 percent, followed by industrial engineering at 23.3 percent and civil engineering at 21.8 percent. Mechanical engineering (11.2 percent) and electrical and electronics engineering (10.1 percent) sit at the bottom. Biomedical and environmental engineering are often named as the most female-dominated on degree data, but both employ too few people for BLS to publish a reliable occupational figure.
Sources
Every figure in this article comes from one of the following. If you are citing a statistic, please cite the original rather than us.
Natalie Kaminski is the CEO and co-founder of JetRockets, a women-owned, Rails-only software agency in New York, and a contributor at InformationWeek on diversity in technology. She is happy to talk to journalists working on this subject.
