How to Break Bad Habits: What the Research Shows

You do not delete a bad habit, you change the response its cue triggers. In a 2011 experiment by Adriaanse and colleagues at Utrecht University, plans written in the negative form, if this situation then I will not do it, produced an ironic rebound and strengthened the habit, while plans naming a replacement behaviour did not. Against habits that are already strong, Quinn, Pascoe, Wood and Neal found in 2010 that removing the trigger did not help and vigilant monitoring did.

Nearly every guide on this subject arrives at the same sentence: do not just stop the bad habit, replace it. The advice is correct, and it is also where most guides stop. What they leave out is that the wording of the replacement decides whether it works. In a published experiment, the most intuitive phrasing, the one that sounds like a firm decision, left people performing the habit more than a comparison group that had made no plan at all.

How do you break a bad habit?

You keep the cue and change the response, and you write that replacement as an instruction to do something, never as an instruction to refrain. This is not a stylistic preference. Marieke Adriaanse and colleagues at Utrecht University tested the two phrasings directly, published the result in Personality and Social Psychology Bulletin in 2011, and found that the refusal version made things worse rather than better. The rest of this article is what the evidence supports underneath that sentence, and where it stops short of what popular advice promises.

That definition is doing real work, because it rules out the mental model most people bring to the problem. You are not removing a file. You are training a second response to win a race that the first response has been winning for years, in a place where it always wins. Here is what has actually been tested, and what each thing was found to do.

Two rows in that table contradict what you have been told. Removing the trigger is the headline advice on most pages about breaking habits, and it was the strategy that failed against the habits people most wanted to break. Planning not to do something is the phrasing almost everyone reaches for, and it was worse than planning nothing.

Why is it so hard to break bad habits?

Because a habit is started by the situation rather than chosen in the moment, so a decision made this morning never reaches the moment the behaviour actually begins. The clearest demonstration is a pair of field experiments by David Neal, Wendy Wood, Mengju Wu and David Kurlander, published in 2011. Habitual cinema popcorn eaters were given popcorn that was fresh or stale. They ate the same amount of both, and how hungry they were and how much they liked the food barely mattered.

The interesting part is the two conditions under which that stopped being true. Habitual eaters ate out of habit only when they were in the cinema, the context associated with past performance. Move the same people to a meeting room and the habit did not run. And in the second study, habitual eaters ate on autopilot only when they could eat with their dominant hand. Asked to use the other hand, they ate according to how much they actually liked the popcorn, like everyone else. Occupying the motor pattern was enough to hand control back to preference.

Established habits do not run on wanting. They run on being in the place where they have always run.

This is why the standard advice to want it more does not do much. People with weaker cinema popcorn habits did eat according to liking and hunger, in every condition. Motivation guides behaviour perfectly well right up until a habit is strong, and then it stops being the thing in charge. If you are wondering why the same resolution fails every January, that is the mechanism, and it is also why habit formation takes a median of 66 days in the other direction.

Why does telling yourself not to do it make it worse?

Because a plan phrased as a refusal has to name the habit in order to forbid it, and naming it in the cue situation is exactly the pairing that keeps the habit alive. Adriaanse, van Oosten, de Ridder, de Wit and Evers tested plans with an explicit if situation, then not habitual response structure, predicting from ironic process theory that they would strengthen the very association they were meant to sever. That is what they found: negation plans produced ironic rebound effects, cognitive and behavioural, relative both to a replacement plan and to holding the intention alone. The effect was most likely to appear when the habit being negated was strong, which is to say precisely when you would reach for a firm plan.

Read the left side of that diagram again, because it explains a frustration most people have felt without being able to name it. To hold the thought not the biscuit you have to hold the thought biscuit. Rehearse that pairing in advance, attached to the exact moment of stress, and you have spent the week practising the association you wanted to weaken. The plan felt like resolve. Mechanically it was repetition.

How do you word a replacement that actually works?

Name one specific action, tie it to the exact moment the habit fires, and make it something you can physically do in that moment instead. This is old ground clinically. Nathan Azrin and Gregory Nunn published habit reversal in Behaviour Research and Therapy in 1973, built on a competing response: a movement performed at the first sign of the urge that is physically incompatible with the habit. Not a resolution to do better, a thing your hands are doing instead.

The format matters more than the content. Compare I will stop scrolling in bed with when I get into bed, I will put the phone on the shelf and open the book on the nightstand. The second names a situation, a place and an action. The first names an outcome and leaves the moment itself unscripted, which means the habit gets to decide it.

Now the honest part, and it is the reason this article does not promise you a fix. A second 2011 paper, by Adriaanse with Peter Gollwitzer, De Ridder, de Wit and Kroese, took apart what a replacement plan is actually doing. In a control condition, on encountering the cue, the habitual response was more mentally accessible than the alternative. A counterhabitual plan removed that advantage. But the alternative did not then become automatic in its place. The authors conclude that such plans increase your flexibility to choose in the critical situation, and that behaviour departs from habit only to the extent that you also hold a strong intention toward the alternative.

The plan does not win the race for you. It takes away the head start, and then something still has to make you choose.

That is a real limitation and most pages skip it, because it makes the advice less tidy. It also points at what to do about it. If the plan levels the moment rather than deciding it, then the question is what supplies a live reason at the cue, week after week, once the initial motivation has flattened out. Willpower is a poor candidate, for the reason the popcorn study made obvious.

When is the easiest time to break a habit?

When the context changes, because a habit that cannot be cued by its usual surroundings is not performed and briefly comes back under intentional control. Wendy Wood, Leona Tam and Melissa Guerrero Witt followed students transferring to a new university and tracked their exercising, newspaper reading and television watching, publishing the results in the Journal of Personality and Social Psychology in 2005. The habits survived the transfer only where aspects of the performance context did not change, for example where a student carried on reading the paper with other people. Where the context went, the habit went with it, and in some cases behaviour returned to being guided by current intentions.

The practical reading is narrow but useful. A move, a new job, a changed commute, a different schedule, a rearranged kitchen: these are the moments when the cue structure is already broken for you and a replacement has an unusually clear run. It is not a fresh start in the motivational sense. It is a period during which the old response is not being triggered, which is a different and better thing. If you have a move coming up, that is when to spend your attention, not on the first of January.

When the context is not going to change, you are back to the strategy Quinn, Pascoe, Wood and Neal actually found working against strong habits. Their two experience-sampling diary studies compared what people reported doing. Stimulus control inhibited responses to temptations, but against strong habits the authors report that no strategy was useful except vigilant monitoring, which they describe as thinking do not do it and watching carefully for slip-ups. A follow-up experiment indicated that monitoring works by heightening inhibitory control rather than by weakening the habit itself.

Note what that means. Monitoring does not dismantle anything. It puts a watcher at the moment of the cue, which is the moment your morning decision could not reach. That is the same gap the replacement plan leaves open, approached from the other side.

What the pages that rank on this leave out

We went through the pages ranking for this question on 28 August 2026 and checked them against the research above. Two of them we could read in full; the table says exactly what we verified on each, and nothing more.

Both pages tell you to replace the habit. Neither tells you that the natural way to write a replacement plan, as a refusal, was tested against the alternative and came out behind making no plan at all. That gap is the reason this article exists. The finding has been in print since 2011, in a mainstream social psychology journal, and it has not reached the advice.

One note on method, because it cuts both ways. Several health publisher pages in this results set could not be read from our server at all; they return a challenge page to anything that is not a browser. We have left them out rather than describe pages we did not read, so treat the table as a sample of the top results and not as a census of them.

We would rather flag a second thing than let it pass. None of the strategies here comes with a number of days attached, and if a page gives you one for breaking a habit, ask where it came from. The best-known figure in this field, a median of 66 days from Lally and colleagues at University College London, measured the growth of automaticity in a new behaviour being formed, not an old one being broken. Those are not the same process run backwards, and there is no equivalent published median for the reverse.

Where a habit tracker fits, and where it does not

A tracker cannot supply the replacement behaviour or the plan. What it can do is cover the two gaps the research keeps pointing at: something has to give you a live reason at the moment of the cue, and something has to make the slip visible while it is still a slip.

StellarHabit is built around the second half of that. You track the replacement behaviour, not the habit you are quitting, which follows the logic of this whole article: the thing worth recording is the response you want the cue to produce. Progress shows as the percentage of days you actually completed rather than a streak, so a single bad Tuesday does not wipe the record and give you a reason to abandon it. We wrote about why streak counters make people quit separately. And the number is visible to the people in your clan, which is the part a private spreadsheet cannot do: in the Harkin meta-analysis of 138 studies, monitoring worked better when progress was reported to someone else than when it was kept private.

Being honest about the limits: this is not treatment, and it will not help with a dependence. It also does nothing for you if you have not done the work above first, because a tracker with stop snacking in it is a negation plan with a checkbox. If you need candidate replacements, our list of 75 bad habits pairs each one with a specific behaviour to swap in, and the guide to building good habits covers how many replacements you can reasonably run at once. The answer is fewer than you think.