Women in STEM: Statistics, Careers and the Gender Gap in 2026

How many women work in STEM in the UK? Which subjects have the largest gender gaps? And is the situation actually improving?

The answer depends considerably on what — and who — we count.

Search for the percentage of women in STEM and you may encounter a surprisingly wide range of answers.

25%.

27%.

Perhaps something else.

At first glance, this looks like a statistical problem. It is actually a definition problem.

There is no single official definition of a STEM occupation in the UK. Different organisations include different professions, use different datasets and measure different populations. Medicine may be included in one definition and excluded from another. A university statistic might describe students rather than workers. An engineering statistic might cover a much narrower group than a figure labelled simply “STEM”.

So before asking how many women are in STEM, there is a more important question:

What exactly do we mean by STEM?

How Many Women Work in STEM in the UK?

There is no single percentage that answers this question for every definition of STEM.

The Office for National Statistics has noted that there is no official definition of which occupations should be classified as STEM jobs. Many occupations use scientific or technical knowledge to different degrees, so the boundary depends partly on the classification being used.

That is why apparently contradictory statistics can all be legitimate.

Department for Education analysis of the UK STEM workforce, for example, found that women represented 25% of the STEM workforce in 2023 under its occupational definition.

Inside that headline number, however, the differences were enormous.

Women represented approximately 8% of engineering technicians, 15% of programmers and software development professionals, and 46% of actuaries, economists and statisticians.

All can appear within a broad discussion of STEM. Their gender composition is completely different.

Other analyses using different definitions have produced different headline percentages for women's representation in STEM.

The useful conclusion is not that one number must be right and the others wrong. It is that “women in STEM” is an umbrella statistic. The interesting information appears when we open the umbrella.

There Is No Single STEM Gender Gap

The same problem appears in higher education.

Women are often described as underrepresented in STEM as though Biology, Computing, Engineering, Mathematics, Medicine, Physics and Veterinary Science were experiencing versions of the same phenomenon.

They are not.

The She Invented That Statistics Observatory uses Office for Students population data to examine ten selected undergraduate subject groups at English higher-education providers.

For 2023–24, female representation was:

Engineering and Technology — 19.1% Computing — 20.1% Mathematical Sciences — 34.5% Physical Sciences — 45.8% Biological and Sport Sciences — 49.3% Geography, Earth and Environmental Studies — 52.6% Medicine and Dentistry — 62.2% Subjects Allied to Medicine — 79.6% Psychology — 82.2% Veterinary Sciences — 84.8%

These figures refer to all undergraduate students and all domiciles within the selected subject categories. In other words, they are not restricted to UK-domiciled students.

Put the disciplines beside one another and the idea of a single STEM gender gap becomes difficult to sustain.

Engineering and Computing remain heavily male. Physical Sciences are much closer to balance. Women form a majority in Medicine and Dentistry and very large majorities in Psychology, Subjects Allied to Medicine and Veterinary Sciences.

Broad statements such as “women remain underrepresented in STEM” therefore describe something real while simultaneously concealing much of what matters.

The acronym stays the same. The populations underneath it do not.

This is also why the familiar metaphor of a STEM “pipeline” has limitations.

The data suggest something closer to a network of branching roads.

Students do not simply enter STEM at one end and either remain or disappear. They choose different subjects, qualifications and careers. A student who is good at Mathematics might choose Engineering, Computing, Economics, Medicine or Mathematics itself. A Physics graduate might enter research, software, finance or teaching.

At each junction, the population changes. And the gender balance can change with it.

Is the STEM Gender Gap Closing?

Again, there is no single answer.

Between 2010–11 and 2023–24, female representation increased in most of the ten subject groups tracked by the She Invented That Observatory.

Some of the largest changes were:

Physical Sciences: +7.2 percentage points Geography, Earth and Environmental Studies: +6.1 points Medicine and Dentistry: +5.6 points Engineering and Technology: +4.7 points Computing: +3.6 points

Engineering moved from 14.4% female to 19.1%. Computing moved from 16.5% to 20.1%. Physical Sciences moved from 38.6% to 45.8%.

Those changes matter. A field can remain substantially imbalanced while still making genuine progress.

But progress has not been universal.

Female representation in Mathematical Sciences fell from 38.7% to 34.5%, a decline of 4.2 percentage points across the period. Subjects Allied to Medicine also declined slightly, although women remained a large majority.

There is therefore no national trajectory in which women are simply and steadily “catching up” across STEM. Different disciplines are travelling in different directions and at different speeds.

Engineering: Where the Path Narrows

Engineering provides one of the clearest examples of why the gender gap cannot simply be explained by saying that girls need more science or mathematics.

In the She Invented That dataset, women represented 19.1% of Engineering and Technology undergraduates at English higher-education providers in 2023–24, up from 14.4% in 2010–11.

The gap begins earlier than university.

Recent EngineeringUK figures show girls account for around half of students taking GCSE Mathematics and Physics.

By A level, the picture has changed substantially. In 2026, female students represented approximately 37% of Mathematics entries and 24% of Physics entries. They were also underrepresented in Further Mathematics, Design and Technology and Computing.

That is a subject-participation gap. But it is not the whole story.

EngineeringUK examined students who had already studied Mathematics and/or Physics at A level.

Among male students with those subjects, 23% progressed into Engineering and Technology undergraduate study. Among female students, only 8% did.

That is a different phenomenon: a conversion gap.

Women can reach the point at which engineering is a realistic educational option and still choose it at a much lower rate than men with similar subject backgrounds.

Even among students who had taken both Mathematics and Physics, the difference remained. EngineeringUK found that 39% of men in this group chose Engineering and Technology compared with 29% of women.

Nor does the imbalance belong solely to universities.

Recent EngineeringUK figures report that girls account for around 12% of Engineering and Technology T Level students, while women represent approximately 20% of engineering and technology apprenticeship starts.

In the workforce, EngineeringUK estimates that women account for around 17% of engineering and technology workers.

Even those figures hide substantial differences between engineering disciplines. Female representation is considerably higher in areas such as Chemical, Process and Energy Engineering than in Mechanical or Production and Manufacturing Engineering.

There is no single point at which women disappear from engineering. The road branches repeatedly — during subject choice, university applications, technical education, apprenticeships and employment.

Computing: Progress, but Still One in Five

Computing tells another version of the story.

In 2023–24, women represented 20.1% of Computing undergraduates in the She Invented That dataset. That was up from 16.5% in 2010–11.

So female representation has increased. But after fourteen academic years, women still represented only around one in five Computing students.

The workforce picture also remains substantially imbalanced.

Department for Education analysis estimated that women represented 15% of programmers and software development professionals in the UK in 2023.

Computing is particularly interesting because its gender composition has not always looked inevitable.

Women played important roles in early programming, software development and information science. Programming itself was not always culturally imagined as overwhelmingly male work.

That history does not prove why today's gender imbalance exists. It does demonstrate something important: the gender composition of a technical field can change.

A profession's current demographic pattern is not necessarily an inherent property of the work itself.

Science and Medicine Tell a Different Story

Move away from Engineering and Computing and the familiar STEM narrative begins to change.

Women represented 45.8% of Physical Sciences students in the She Invented That dataset in 2023–24 and 49.3% of Biological and Sport Sciences students.

In Geography, Earth and Environmental Studies, women were already a slight majority at 52.6%.

Then the balance moves further.

Women represented 62.2% of Medicine and Dentistry students, 79.6% of students in Subjects Allied to Medicine, 82.2% in Psychology and 84.8% in Veterinary Sciences.

This does not mean gender has ceased to matter in these professions. Student participation is only one stage of a career.

But it does mean that a policy discussion about women in STEM cannot begin with the assumption that every scientific discipline is struggling to attract women in the same way.

Some fields face a severe entry problem. Others do not. And in some, the more important question may emerge much later.

Getting In Is Not the Same as Staying In

This distinction becomes particularly important in research careers.

The UK's 2026 Women in Research Charter reports that women represent 53% of students in science-related undergraduate subjects.

At junior and mid-level research positions, women account for approximately 51%. Further up the academic hierarchy, representation declines.

Women account for around 44% of senior academic leadership and 31% of professors.

The problem has changed. At undergraduate level, the question may be who enters a subject. At senior research level, it becomes who progresses and remains.

The Charter also identifies a widening difference in career progression from around age 35, coinciding with a period when caring responsibilities and part-time work become particularly important.

Another government analysis provides a striking illustration of what can happen outside the conventional career path.

An evaluation of the STEM ReCharge programme estimated that there were approximately 58,100 potential STEM returners in the UK in 2023: working-age people who had previously worked in STEM, were economically inactive because of caring responsibilities and expected that they might work again.

Around 71% were women.

Almost three quarters lived in households with dependent children, and around one third had been out of work for five years or more.

These are people who already crossed the supposed entry barrier. They studied, trained or worked in STEM.

The issue is no longer attracting them to the field. It is enabling them to return.

That requires a very different intervention from persuading a teenager to study Physics.

Why Do the Pathways Diverge?

Data describe outcomes, not motivation.

Statistics can trace where pathways diverge — between GCSE and A-level Physics, between Mathematics and an engineering degree, or between mid-career research and senior academia — but they cannot single-handedly explain why individuals choose one route over another.

That distinction is important.

Participation data alone cannot prove that stereotypes caused a particular student to reject Engineering, that caring responsibilities caused a researcher to leave academia, or that one outreach programme changed someone's career.

But other evidence can help identify the environment in which those decisions occur.

EngineeringUK, for example, reports a substantial identity gap around engineering.

Only 12% of girls said that being an engineer fitted well with who they were, compared with 38% of boys. Just 16% of girls thought engineering was suitable for them, compared with 44% of boys.

Those figures do not prove causation. They do tell us that girls and boys can arrive at educational decisions with very different perceptions of the same profession.

Elsewhere, the relevant pressures may be different. Entry requirements matter. Knowledge of careers matters. Workplace structures matter. Progression opportunities matter. Caring responsibilities can matter.

The mistake is not considering these factors. It is assuming that the same explanation applies equally across every discipline and every career stage.

STEM Careers Are Not One Career

The phrase “STEM careers” can itself obscure almost as much as it explains.

STEM includes software developers and structural engineers, doctors and data scientists, laboratory researchers and environmental scientists, statisticians and aerospace engineers.

It includes people developing pharmaceuticals, designing renewable-energy systems, studying galaxies, analysing genomes, building robots and teaching Mathematics.

Someone interested in climate change might become an environmental scientist, energy engineer, atmospheric physicist, materials researcher or data analyst.

Someone interested in healthcare might become a doctor — but could also design medical devices, analyse genomic data, develop pharmaceuticals or work in biomedical engineering.

Someone strong in Mathematics might enter engineering, artificial intelligence, statistics, economics, finance, Physics or computing.

This matters because encouraging a young person to “consider STEM” is not yet meaningful career guidance.

The useful questions are more specific: What kind of problems do they want to solve? What skills does the work require? Which qualifications lead there? What does the job actually involve? Where are the opportunities? And how is the profession changing?

Demand for scientific and technical skills remains substantial.

Department for Education analysis found employment in occupations classified as STEM under its definition increased from approximately 7.7 million in 2013 to 9.4 million in 2023 — growth of roughly 22% over the decade.

But even that figure should not be interpreted as a promise that every STEM occupation will grow equally or that any STEM qualification automatically guarantees employment.

Once again, the acronym is less informative than what lies beneath it.

What Should Change?

If there is no single STEM gender gap, there cannot be one universal solution.

Engineering may require interventions around school subject choices, perceptions of the profession and the conversion of mathematically qualified women into engineering education.

Computing may require a different combination of educational experiences, cultural expectations and routes into technical careers.

Academic research may require greater attention to retention, career progression, flexible working and return routes after caring-related breaks.

And disciplines already dominated by women raise different questions again.

This suggests a relatively simple principle:

Measure the discipline. Identify the decision point. Then design the intervention.

That is considerably more useful than beginning with the statement “women are underrepresented in STEM” and assuming that encouraging more girls into STEM is therefore the answer.

Even well-intentioned outreach needs the same precision.

If 200 girls attend an engineering workshop, attendance alone tells us that 200 girls attended an engineering workshop.

Did their understanding of engineering change? Did more choose relevant subjects? Did they apply for engineering courses or apprenticeships? Did their interest persist? Did the programme affect different groups of students differently?

Without those questions, it is easy to measure activity rather than impact.

What the Numbers Actually Tell Us

Statistics are useful because they reveal patterns that broad narratives can hide.

They show that women represent roughly one fifth of Engineering and Computing undergraduates in the She Invented That dataset. They also show that Veterinary Sciences and Psychology look almost like mirror images of those fields.

They reveal substantial progress in Physical Sciences and slower progress in Engineering and Computing. They show that participation can change between school and university. And research-career data demonstrate that entering STEM is not the same as remaining and progressing within it.

None of this means that every discipline must contain precisely equal numbers of men and women.

People choose subjects and careers for different reasons, and statistics cannot tell an individual what she ought to study. But they can help identify where choices begin to narrow in systematically different ways.

The evidence does not reveal one STEM gender gap waiting for one solution.

It reveals a series of different imbalances appearing at different points: subject choice, university entry, conversion into particular disciplines, technical training, workforce participation, career breaks and senior progression.

The useful question is therefore not simply how to get more women into STEM. It is where the pathways diverge, for whom, and at what stage.

Only then can we distinguish a headline statistic from a problem that can actually be solved.

Explore Women in STEM Data

Use the She Invented That Statistics Observatory to explore fourteen years of female student numbers and representation across ten selected undergraduate subject groups at English higher-education providers.

Compare Computing, Engineering and Technology, Mathematical Sciences, Physical Sciences, Medicine and Dentistry, Veterinary Sciences and other disciplines from 2010–11 to 2023–24.

Then continue with The STEM Gender Gap Isn't Where You Think It Is, Engineering's Missing Half and When Did Computing Become a Man's World? for deeper investigations into individual parts of the data.

Find Opportunities for Women in STEM

She Invented That also tracks selected scholarships, awards, grants and programmes relevant to women in STEM.

Browse current opportunities and follow the routes connecting education, research and STEM careers.

Method Note

There is no universally accepted definition of STEM used across every UK dataset. Workforce estimates cited in this article may therefore describe different occupational populations.

The She Invented That Statistics Observatory uses Office for Students population data for students at English higher-education providers.

The figures presented here refer to all undergraduate students and all domiciles within the selected subject categories unless otherwise stated. They should therefore not be interpreted as statistics for UK-domiciled students alone.

The ten undergraduate subject groups presented by She Invented That are a selected set used for the Observatory. They are not an official Office for Students definition of STEM.

Student population statistics describe participation and representation. They do not, by themselves, establish why students choose particular subjects or careers.

Statistics from EngineeringUK, Department for Education and other sources describe different populations and should not be treated as one continuous statistical series.