Will disclosing hurt my career?

What the population data does and doesn’t show

Vincent and Ralston analysed pooled UK graduate-outcomes data from 2012 to 2018, drawn from two cross-sectional surveys run by the Higher Education Statistics Agency. The full dataset held 1,326,416 graduates who reported their outcomes, of whom 5,646, or 0.4%, reported social communication or autistic spectrum disorder. Autistic graduates recorded lower full-time employment and less favourable income distributions than the comparison groups, and their degree subjects were more varied than stereotypes suggest.

Three limits matter more than the headline number.

The autistic group is 5,646, not 1.3 million. The large figure is the comparison dataset. Every finding about autistic graduates rests on the smaller group, and the income findings rest on less than that: information was available for only 1,095 autistic graduates in full-time work, which is why the authors say income conclusions must be drawn tentatively.

Disability status was self-reported. The authors note there was no diagnostic information and no obligation on anyone to disclose. So the autistic group is defined by what people told a survey, which on a page about disclosure is worth sitting with rather than skipping past.

It’s observational. These are measured group disparities. Nothing in the design identifies what caused them, and nothing in it compares people who disclosed at work with people who didn’t. It gets cited as though it does.

What’s known about disclosure specifically

Davies, Romualdez, Pellicano and Remington published a scoping review in 2024 covering 33 studies. Its most important finding, for this question, is about the literature rather than about disclosure: they were unable to identify any studies that directly focused on career progression. Most eligible studies examined employment experiences more broadly, and the reviewers extracted the parts that seemed relevant.

That matters because it sets a ceiling on what follows. Nothing below is a study designed to answer this question.

Within that wider literature, the review separates two things that are easy to run together. Participants in six studies raised concerns about the potential negative impact of disclosure on their career success: being pigeon-holed into autism-specific work, worries about job prospects, a fear that a diagnosis might be used as a reason not to promote someone. Those are anticipated harms.

Participants in three further studies gave qualitative accounts of reduced opportunities following disclosure. One reported being sacked after receiving their diagnosis. One was turned down for an internal opportunity after disclosing. A third described an openly autistic acquaintance being given fewer opportunities, which is a witnessed account rather than the participant’s own.

So: reported experiences, attributed by the people describing them to disclosure, in studies not built to test that. That documents a plausible risk. It doesn’t isolate disclosure as the cause, and it can’t say how often the outcome occurs. Anyone quoting a percentage is going beyond the source.

The review also found underemployment common, gaps in employment history making progression harder, and employment support the most frequently discussed facilitator, though little research has evaluated it.

A separate 2024 scoping review looked at disclosing non-visible disabilities in educational workplaces and reached a compatible conclusion from another angle: the reasons behind disclosure decisions are complex and emotive, both individual and environmental factors shape them, and stigma and perceived discrimination are key barriers while personal agency is a key enabler. It covers educational workplaces specifically, so it travels only so far.

The hiring experiments, and what they can’t tell you

Here the evidence gets more precise and, in an important way, weaker.

May and colleagues ran a study with three conditions. Participants watched mock job interviews with autistic and non-autistic candidates, rated them on social dimensions, and said how likely they’d be to hire each one. In the disclosure-only condition, where participants were told which interviewees were autistic, autistic candidates were rated less favourably and were less likely to be hired. In the training-only condition, where participants completed neurodiversity training but weren’t told who was autistic, the same thing happened. Only in the training plus disclosure condition were autistic candidates as likely to be hired as non-autistic candidates, even though they still received less favourable ratings on some social dimensions.

A follow-up by an overlapping team tested whether that effect lasted. Participants, drawn from undergraduates and the general US population via Prolific, completed neurodiversity training either two weeks or two months before reviewing taped interviews. At both delays, the difference in stated hiring likelihood between autistic and non-autistic candidates was not significant, while autistic candidates were still rated as more awkward and less similar.

Read together, the experiments suggest disclosure alone did not close the hiring-intention gap; the condition in which autistic and non-autistic candidates did not differ significantly on stated likelihood of hire was disclosure combined with neurodiversity training.

Two pieces of precision matter there. There was no no-disclosure control condition, so the design can’t show that disclosing made outcomes worse than staying quiet, only that the gap persisted when disclosure stood alone. And a non-significant difference is not an equivalence test: “no significant difference in stated hiring likelihood” is not the same as a demonstrated absence of bias.

Now the wider limits, which are serious. These are analogue studies. The participants were students and online panel respondents watching recordings, not employers filling real vacancies with real budgets and real consequences. What was measured is a stated hiring intention in a low-stakes setting. That’s a legitimate way to isolate an experimental manipulation, and a poor guide to what a hiring manager does in an actual process. Both papers also come from one research programme with overlapping authors, so they aren’t independent replications.

What the experiments show is that evaluator knowledge changed how disclosure landed in a lab. Whether that transfers to real hiring, let alone to a career, is untested.

What the ADHD evidence says

The literature above is almost entirely about autism, but the ADHD side isn’t empty.

McIntosh and colleagues surveyed 166 working adults with ADHD in 2022 and examined proactive disclosure directly, meaning telling an employer before any performance problem arises. Stigma consciousness predicted less proactive disclosure; psychological safety predicted more. Proactive disclosure was associated with thriving at work, and fully mediated the relationship between stigma consciousness and thriving in the authors’ path model.

Every variable there was measured at one time point, through self-report, in an online panel sample. So it cannot show that disclosure caused thriving, and the causation could as easily run the other way, with people who are already thriving finding it safer to disclose. It also says nothing about promotion, pay or retention, which is what this page’s question is usually about.

More broadly, Lauder, McDowall and Tenenbaum synthesised 143 studies in 2022 and found limited research on the effectiveness of workplace interventions for adults with ADHD, with much of the existing work evaluating pharmacological interventions that are hard to transfer to a workplace context.

So ADHD readers are getting a thinner literature, not an absent one, and the one study pointing directly at disclosure points somewhere more hopeful than the autism material does.

What the accommodations side shows

If the question underneath yours is whether disclosure is worth it, the other half of the ledger is what disclosure can unlock, and there’s a review of that too.

A 2025 systematic review of workplace accommodations for autistic employees synthesised ten empirical studies published between 2010 and 2025, spanning randomised trials, cohort studies, surveys and qualitative work. Its authors judged meta-analysis unfeasible given the heterogeneity and used a narrative synthesis instead, so it reports directions rather than pooled effects. It grouped what gets used into four categories: assistive technology; organisational, sensory and environmental adaptations; supervisory and relational support; and skills and psychosocial support.

Its overall finding was that accommodations were linked to better job acquisition, stability, satisfaction and productivity. Its qualifications were substantial: most included studies had small samples and relied on self-report, the methods were heterogeneous, and broader autism groups were excluded, which limits how far the findings generalise. It’s a single-author review of ten studies, so it establishes a direction rather than a magnitude.

The part that speaks to this page is what it says about variation. Effectiveness varied by relational quality, disclosure practice and organisational culture, and the review named stigma, lack of knowledge and uneven policy implementation as the barriers that persisted. Two of its four categories, supervisory support and relational support, aren’t things you can obtain by producing a document. They depend on the person you report to.

It isn’t one decision

Framing this as a single irreversible act makes it harder than it is, and the framing is usually wrong.

Telling one colleague you trust, asking for a specific change without naming a reason, connecting a change to a medical condition without naming a diagnosis, and making a formal declaration to an organisation are four different acts. They carry different risks, produce different records, and can happen years apart or not at all. The evidence on this page is about the formal end of that range, and much of what people want from disclosure sits at the informal end.

The scoping review on non-visible disabilities found decisions turned on whether, to whom, how and when, all four at once. That’s a more accurate picture of the decision than yes or no, and it’s the one worth bringing to it.

What seems to move it

Pulling the threads together, the evidence points at the environment more consistently than at the decision.

The hiring experiments found the evaluator’s knowledge changed the outcome. The accommodations literature found effectiveness varied by relational quality and organisational culture. The disclosure review found stigma and perceived discrimination were the main barriers. The career progression review found the support that might help exists but is barely evaluated.

None of that tells you to disclose or not to. What it suggests is that the question “will this hurt my career” may be less answerable than “what is this particular workplace like, and who would I be telling.” A place that has trained its managers and a place that hasn’t are, on this evidence, different propositions wearing the same job title.

This page is about measured career outcomes and their limits. If what you’re weighing is the broader decision, with its mixed lived outcomes and the factors that shift them, that’s the subject of should I disclose autism at work.

Worth remembering too that this isn’t one irreversible act. Telling one trusted colleague, requesting a specific change without naming anything, and making a formal declaration to an organisation are separate decisions with separate consequences, and you’re not obliged to make them all at once or in any particular order.

What none of these studies can do is tell you your odds. The disparities are measured, the harm is documented in reported experience, and the frequency and the cause are both unknown. In the analogue hiring studies, evaluator training changed how disclosure landed; whether that translates into real career outcomes is untested. That’s a thinner conclusion than the question deserves, and it’s the one the evidence supports.