Nobody asks this question well rested. That is the first useful fact about it, and the research agrees. In 2024, Samira Feizi and colleagues surveyed 2,486 doctoral students across twenty Canadian universities and mapped the route by which the quit-thought arrives [1]. It does not travel directly from the science to the decision. It travels through the body, in a chain, and the chain works like a row of dominoes. Stress falls first. Falling stress knocks over emotional and social wellbeing, sleep and mood and the sense of belonging somewhere. Falling wellbeing knocks over satisfaction with the programme, which is where the doctorate itself starts to look like the problem. And only at the end of that row does the last domino fall: the intention to quit. The shape of that chain is the whole problem with the question. By the time “should I quit my PhD” is in the search bar, stress has already had three intermediaries between itself and the thought, and the thought presents itself as being about the doctorate. Sometimes it is. Often it is a status report from further up the chain, wearing the doctorate’s clothes.
Table Of Contents:
- Elena types the question at twenty to midnight
- Case one: exhaustion wearing doubt’s clothing
- Case two: the doctorate is fine, its conditions are not
- Case three: the direction has changed, and leaving is the right call
- The question underneath the question
- The protocol: how to actually decide
- A word for the supervisor who has heard the quit-talk
- Elena, two weeks later
- Frequently asked questions
- References
Which is why this article will not tell you to stay, and will not tell you to leave. If you arrived hoping for the definitive guide to quitting your PhD, I have to disappoint you: there is no such guide, and anyone selling one has not met enough doctorates. What I can do is lay out the three situations hiding inside the question, give you a test for each, and help you work out which one is actually yours, because they have three different answers and the whole difficulty is that they feel identical from the inside.
Elena types the question at twenty to midnight
Elena is in the third year of her PhD in developmental biology. Her results are fine. Not spectacular, fine: a first-author paper in revision, a method that works, a supervisor who is neither a mentor nor a menace. And at twenty to midnight on a Tuesday she is holding her phone in the dark, typing “should I quit my phd” into a search bar, then deleting it before pressing enter, because typing it felt like being caught. She has done this four times this month. In the morning, at the bench, the question feels absurd, and she files it under bad night. By the following Sunday evening it is back, and it feels like the only honest question she has.
If you recognise the deleting-before-enter moment, this article was written for you. And the first thing to establish is which of three situations Elena, or you, is actually in, because the situations look identical from the inside at midnight and could not be more different in what they call for.
Case one: exhaustion wearing doubt’s clothing
The first situation is the one the Feizi chain describes, and in my experience it is the most common of the three; the distress numbers in the doctoral literature point the same way, though nobody has counted my three cases, because the taxonomy is mine. The strain of the work has run down the systems the work depends on, sleep first, then composure, then attention, and the depleted state generates doubt as a symptom, the way a fever generates chills. The mechanics of that rundown, strain compounding like debt with the interest paid in sleep and attention, are the subject of my article on PhD burnout. What matters here is the diagnostic tell, and it is the pattern you have already noticed: the doubt tracks energy, not evidence. It is loudest on Sunday nights and at midnight, quietest mid-morning after real sleep. Nothing about the doctorate changes between Tuesday midnight and Wednesday morning. The battery does.
Here is the test in behavioural terms. Log the quit-thought for two weeks, nothing more: each time it arrives, note the day, the time, and the hours of sleep behind you. If the thought clusters at the bottom of your energy curve, you are not looking at a verdict on your doctorate. You are looking at a gauge reading on yourself. That is not a reason to dismiss it. A gauge reading is information, and the February 2026 cohort we measured shows what it is information about: researchers arriving with 69% in the at-risk range for anxiety screening, and eight weeks later, with working hours statistically unchanged, that share at 34% and self-rated workday performance up from 6.2 to 7.4. The doubt that dissolves when the exhaustion is treated was never about the science. The full measured results are in my article on whether resilience training works.
I hold a strong opinion here and I own it: no irreversible decision should be made at the bottom of the energy curve. Not because the tired mind is stupid, it is not, but because it is running on the wrong data. Sleep-deprived brains overestimate their coping and underestimate their options; the same distortion that makes sixty-hour weeks feel sustainable makes doctorates feel unfinishable.
Case two: the doctorate is fine, its conditions are not
The second situation produces the same midnight search and has a different anatomy. The energy log comes back flat: the thought does not track exhaustion, it tracks contact with one specific part of the doctorate. The supervision meeting, and only the supervision meeting. The project that died in month eighteen and is being dragged onward for reasons nobody will say out loud. The lab whose culture runs on visible suffering. When Castelló and colleagues studied why doctoral students consider leaving, the answers were rarely “the research” [2]. They were institutional and personal conditions around the research: supervision, integration into a scholarly community, the structure of the programme itself.
This case has its own test, and it is a single question: if the one thing changed, would you stay? Be concrete about the one thing. If your honest answer is “if I had a supervisor who read my drafts, I would want the next three years”, then you are not looking at a quit decision. You are looking at a repair decision that has been misfiled as a quit decision, because repair feels impossible and leaving feels available. Repair is usually less impossible than it feels at midnight: supervision relationships can be restructured, committees exist for just this situation, projects can be descoped, and in more cases than most students believe, supervisors can be changed. I walk through the repair side in my article on the PhD supervisor relationship. Leaving a workable doctorate to escape a fixable condition is paying for a renovation with the house.
Case three: the direction has changed, and leaving is the right call
The third situation is real, and pretending otherwise would make this article marketing, which you were trained to distrust. Some people mid-doctorate discover, with a clear head and a full battery, that the destination is wrong. Not the workload, the destination. The tell is the opposite of case one: the thought does not cluster at midnight, it survives the morning. And it is usually a pull, not only a push: there is something specific on the other side, a field, a role, a kind of work, that has been getting more vivid while the doctorate has been getting more procedural.
And here the mind deploys its most respectable-sounding bad argument: “I cannot quit now, I have already put three years in.” Sit with what that sentence actually claims, because economists have a name for it. Hal Arkes and Catherine Blumer showed across a series of experiments that people continue endeavours they would never newly choose, in proportion to what they have already spent on them [3]. The sunk cost fallacy, in doctoral form, runs: the three years are spent whether you stay or leave, they are not recoverable in either direction, and so they are not an argument for either direction. The only question a clear head is entitled to ask is forward-looking: given everything I now know, is the next three years of this the best use of the next three years of me? Notice the trap in full: “I cannot have wasted three years” is a sentence that, obeyed, purchases a fourth.
Two honesties are owed here, one on each side. First, sunk cost reasoning does not mean leaving is always right; it means the past is the wrong variable. If you are eight months from submission, the remaining cost is small and the completed degree has real option value, and a forward-looking calculation can favour finishing, not because of the years behind but because of the months ahead. Second, the sharpest point this article has to make, so I will make it plainly: the most consequential mistake available here is not staying too long or leaving too soon. It is exiting a case-one situation while believing it is case three. Exhaustion writes very convincing resignation letters, and the people who sign them are usually the ones who never ran the timing log.
The question underneath the question
Everything so far has been mechanics: energy, conditions, direction. There is a fourth force, and it does not behave like a case. It cuts across all three, which is why it gets its own section instead of a number.
Most researchers can recite the reasoned chain for doing their doctorate, the career logic, rehearsed so often it runs without thinking. For some, though, the doctorate has quietly taken on a second job that appears nowhere in that chain: proving something inward. That you are clever enough. That you are worth the investment. That the sentence installed somewhere in childhood, the one that says not good enough, is wrong. The joy of conducting research is still there on the surface. Underneath it, the intellect is running a second project: an argument with a belief about yourself. The argument is subconscious, which is what makes it tireless. I cannot tell you how common this is; nobody can measure a motive that hides from its owner. I can tell you that I meet it in our trainings often enough to name it.
My own doctorate began as a ticket: a proof point to open doors into the wide world outside academia. Somewhere in the middle it morphed into something else. The supervisor was difficult. I did not like my project, which had been swapped for another between my committing to the lab and my first day. The friendships among my colleagues were real, and we were in it together, the long hours, the constant failure, the scarce recognition. But the contest that actually ran my doctorate was none of that. It was me against me. What carried me through was a mantra installed in childhood: I can do anything if I put my mind to it; the only question is how painful it is going to be. I finished. It took me years to see the mantra for what it was: not a plan, a core belief with a work ethic.
Here is why this force matters for your decision, whichever case you are in. When the doctorate is proving apparatus, neither staying nor leaving answers the question, because the question was never about the doctorate. Finish as proof, and the proof does not arrive: the day after the defence feels strangely like the day before it, a mechanism Paul takes apart in his essay on the arrival fallacy. Leave as escape, and the belief is portable: change the arena and it re-presents, with the same terms. The beliefs doing this work are formed early and operate automatically, and unpicking them is not a productivity problem; it is the territory Paul works in one-to-one, and his essay on the gifted child is where I would send anyone who read the mantra paragraph above with a jolt of recognition.
This force has a test too, and it is the shortest one on this page. Ask yourself, once, answered without an audience: if nobody would ever know whether you finished, would you still want to? If the answer arrives quickly, in either direction, the doctorate is probably just a doctorate, and the three cases above will settle your question. If the question lands with a thud, sit with the thud before you run any protocol, because a doctorate cannot win an argument with a core belief. It was never the right instrument for that job, and neither is quitting one.
The protocol: how to actually decide
Put the three tests in sequence and the question becomes workable. This is the audit, and it costs two weeks and no commitment.
Step one, run the timing log. Two weeks, every occurrence of the quit-thought, with time of day and sleep behind it. Clustered at the energy trough: treat case one first, decide nothing yet. The doubt gets its hearing after the battery does, because only then is it giving evidence rather than symptoms.
Step two, run the one-thing test. Name the single condition which, changed, would make you want to stay. If you can name it, you are in case two: the next step is a repair conversation, not a resignation, and it is worth one serious attempt before any exit. If the sentence “if X changed I would stay” has no true value of X, continue.
Step three, run the morning test, and only now do the arithmetic. Does the wish to leave survive a rested morning, and is there a pull on the other side you can describe in a sentence to another person without flinching? Then do the forward-looking sum with the sunk years deliberately excluded: remaining cost against remaining value, including the option value of a finish that may be closer than it feels. If leaving still wins on that arithmetic, leaving is not failure. It is the correct output of a well-run decision, and the doctorate you leave with a clear head will have taught you more than the search bar at midnight ever suspected.
And underneath all three steps, the nobody-would-know question from the section above, asked before the arithmetic is trusted. A thud there outranks the protocol, and the mechanics can wait a week.
One of our Fast Forward participants, who arrived at the training with the quit question already open, put the sequencing better than I can: “If you want to quit your PhD, first take the chance of the course and afterwards quit, if you really still want to. I changed my mind and I will not quit… I will learn and manage.” That is the protocol in one sentence of feedback: exhaust case one before you act on case three. She turned out to be in case one. Some people run the same sequence and turn out to be in case three, and the sequence has served them just as well, because they leave knowing, instead of leaving wondering.
A word for the supervisor who has heard the quit-talk
If you supervise or coordinate, and someone has said the word “quitting” to you, take a straight look at your reflex, because both standard reflexes are wrong. Talking them into staying treats every case as case two. Respecting their autonomy straight out of the door treats every case as case three. The useful response is the diagnostic one: which case is this? Ask about sleep and the shape of their weeks before you ask about their commitment. The doctoral distress numbers, one in two students affected in Levecque’s landmark study [4], say that a large share of the quit-talk reaching you is case one, and case one is trainable: the skills underneath it, workload design, recovery and energy allocation, are what Fast Forward, our project management and personal productivity training, builds in seven weeks. The researcher who was allowed to treat exhaustion before deciding, and then decided, made a decision. The one who was talked into staying, or waved through the door, had a reflex applied to them.
Elena, two weeks later
Elena ran the log, because it asked less of her than the question did: no decision, just fourteen days of noticing. The pattern was not subtle. Eleven entries, nine of them after ten at night or on Sunday evening, every one on short sleep. Her mid-morning self, the one at the bench with the revision open, had not voted to quit once. Then the one-thing test, which surprised her more: her honest sentence was “if my Thursday meeting stopped being an ambush, I would want the next two years”, which meant the midnight verdicts had been naming the wrong defendant. The nobody-would-know question arrived quickly, and the answer was yes, she still wanted the work. So Elena is not quitting, and she has also not “decided to stay”, because staying was never really the question. She is treating the exhaustion, and she is preparing one structured conversation about Thursdays. The search bar at midnight got one thing right: something did need to change. It just was not the doctorate.
Some readers will run the same fortnight and watch it point the other way: flat log, no fixable Thursday, a pull elsewhere that survives every morning. That reader is not Elena, and this page has served them just as well, because they will leave knowing, instead of leaving wondering.
Frequently asked questions
Is it normal to want to quit my PhD?
So normal that the research treats it as a measurable stage rather than a scandal. In a 2024 survey of nearly 2,500 doctoral students, the intention to quit rose predictably as stress wore down wellbeing and programme satisfaction, and a landmark Belgian study found one in two PhD students affected by psychological distress. Here is what nobody tells you: the frequency of the thought says almost nothing about whether acting on it is right. The pattern of the thought does. If it arrives at midnight and dissolves by mid-morning, you are reading a fuel gauge, not a verdict.
How do I know if I should quit my PhD?
Run three tests in order, and do not skip the first. Log the quit-thought for two weeks against your sleep and energy: if it clusters at the trough, treat the exhaustion before you treat the question as real. Then name the one condition which, changed, would make you stay: if it exists, you have a repair problem, not a quit problem, and supervision and project structures are more fixable than they feel. Only if the wish survives a rested morning, with something concrete pulling you elsewhere, does it become a genuine decision, and then you make it on forward-looking arithmetic with the years already spent deliberately left out of it.
Is quitting a PhD a failure?
Only if you define failure as updating a plan with better information, in which case every good researcher fails constantly. The honest distinction is not between finishing and leaving, it is between deciding and reacting. Leaving with a clear head, after the exhaustion has been treated and the fixable has been tried, is a decision, and people who make it tend to describe it later as a good one. Leaving at the bottom of the energy curve, with the sunk-cost argument shouting in both directions, is a reaction, and it is the exit people report regretting, whichever way it went.
Should the years I have already invested stop me from quitting?
No, and the economists have been blunt about this since the 1980s: the work of Hal Arkes and Catherine Blumer on the sunk cost fallacy showed how reliably people continue things they would never newly choose, in proportion to what they have already spent. The three years are gone whether you stay or leave, so they belong in neither pan of the scale. What does belong: the remaining cost of finishing against the remaining value of the degree, including the real option value of a submission that may be closer than midnight makes it feel. Weigh the road ahead, not the road behind.
Why do I want to quit my PhD when my results are fine?
Because the doubt is rarely reading your data; it is reading you. Research on doctoral students, including a 2024 survey of nearly 2,500 of them, shows the quit-thought arriving through a chain that starts with stress and runs through sleep, mood and belonging before it ever mentions the science, which is why it can coexist with a paper in revision and a method that works. And when even that chain does not fit, look at what the doctorate is being asked to prove. A degree doing double duty as evidence of your worth generates doubt no result can settle, because no result was ever going to. Fine results with persistent doubt is not a contradiction. It is a diagnostic.
Should I take a break before quitting my PhD?
Before any irreversible decision, restore the equipment making it. That can be leave, and in many systems a formal interruption of studies exists for this reason, but a break alone often returns you, rested, to the same conditions that ran you down. The stronger sequence is recovery plus the two-week timing log plus one serious repair attempt at whatever is fixable. One of our training participants, who arrived intending to quit, gave the sequencing that I would give: take the chance to rebuild how you work first, and afterwards quit, if you really still want to. She stayed. The point is not that you will. The point is that either way, you will know.
References
- Feizi, S., Knäuper, B., & Elgar, F. (2024). Perceived stress and well-being in doctoral students: effects on program satisfaction and intention to quit. Higher Education Research & Development, 43(2). ↑
- Castelló, M., Pardo, M., Sala-Bubaré, A., & Suñe-Soler, N. (2017). Why do students consider dropping out of doctoral degrees? Institutional and personal factors. Higher Education, 74(6), 1053–1068. ↑
- Arkes, H. R., & Blumer, C. (1985). The psychology of sunk cost. Organizational Behavior and Human Decision Processes, 35(1), 124–140. ↑
- Levecque, K., Anseel, F., De Beuckelaer, A., Van der Heyden, J., & Gisle, L. (2017). Work organization and mental health problems in PhD students. Research Policy, 46(4), 868–879. ↑
Training that treats case one before it becomes case three: Fast Forward builds workload design, recovery and energy allocation in seven weeks, and the Resilience Training measures what changes at both ends.
Talk to us about training for your researchers
Last Updated on August 17, 2026 by Dr Nadine Sinclair
Dr Nadine Sinclair
Dr. Nadine Sinclair is a molecular biologist, strategy consultant and co-founder of Mind Matters. A scientist by training and at heart, she conducted her doctoral research at the Max Planck Institute for Biophysical Chemistry. She has 30,000+ hours of hands-on project management experience, built across 18 years as a strategy consultant at McKinsey and as an independent consultant, working with research institutions, foundations, pharmaceutical and biotech companies and governments. She is the co-creator of the Personal Resilience Indicator and the author of On Track. Since 2018, over 1,500 researchers from more than 30 institutions have trained with Mind Matters.