Why most diets fail (and what doesn’t).

The 83% number

In 2007, a research team at UCLA led by Traci Mann published a review in the journal American Psychologist. They looked at 31 long-term studies of dieters — people followed for two to five years after starting a diet. Among those followed for at least two years, 83% regained more weight than they had lost.

Read that again. Not 30%. Not half. Eighty-three percent ended up heavier than they started.Mann’s group concluded that the data did not support recommending diets as a treatment for obesity — not because diets don’t produce short-term weight loss (they obviously do), but because the long-term outcome for most people is worse than not dieting at all.

That review is almost twenty years old now, and you’ll find researchers who debate the methodology. But the headline finding — that long-term diet adherence collapses for most people, and that regain is the rule rather than the exception — has been replicated across newer reviews of weight regain and adherence. It’s not one paper. It’s a pattern.

A diet you can do for six weeks isn’t a diet. It’s a sprint. And nobody finishes a marathon at sprint pace.

Why it isn’t willpower

The first instinct, when you find out the failure rate is that high, is to assume something is wrong with you for being in the 83%. That’s the wrong instinct. There are two reasons most people regain weight, and neither is moral:

Your body fights back. Roughly 36 weeks after the end of a weight-loss intervention, studies start to see weight variation reduce and a slow, steady regain begin. The mechanisms aren’t mysterious — your resting metabolic rate drops, your hunger hormones shift upward, your reward sensitivity to food increases. The body, in short, defends its setpoint. Aggressively. Some people defend less than others — there’s real individual variation here — but the average effect is real and consistent.

Adherence collapses, predictably. The same body of research finds that self-reported full adherence to a prescribed diet drops from 81% at month 1 to 57% at month 24. That curve looks roughly the same across diet types. Whether it’s keto, low-fat, Mediterranean, or paleo, the line going down is structurally similar.

The reviews that compare diets head-to-head all converge on the same boring, important fact: the predictor of success isn’t the diet you choose. It’s whether you can keep doing it.

The one thing that has held up, long-term

There is one intervention with strong long-term evidence — strong in the way the rest of the field is not. It’s called the Diabetes Prevention Program, an NIH trial that started in the 1990s. They took 3,234 adults who were already at high risk for type 2 diabetes and randomized them into three groups: a placebo, the diabetes drug metformin, and a lifestyle program.

The lifestyle program was unimpressive on paper. It asked for:

• A modest weight loss of 7%. For someone weighing 180 lbs, about 13 lbs.
• 150 minutes a week of moderate physical activity. Walking counts.
• Sustained behavior support — a 16-lesson curriculum and ongoing contact with a coach.

In that population, the lifestyle program reduced diabetes incidence by 58%. Metformin, the drug, reduced it by 31%. Then they kept following these people. At 22 years — 22 yearsafter the original intervention — the people who’d been in the lifestyle group were still 25% less likely to have developed diabetes than the placebo group. That’s the kind of long-term effect that almost never shows up in the diet literature.

Two honest things about this. First, the trial was specifically in high-risk pre-diabetic adults, so the headline numbers are about that group, not a promise for everyone. Second, the framework— modest changes, ongoing support, no end date — is what generalizes. The intervention wasn’t aggressive. It was modest, supported, and structurally sustainable. That’s the part of the story most fitness marketing leaves out.

The crash-diet failure modes, briefly

All of these produce fast initial weight loss. None of them have outcomes past 12 months that look meaningfully different from doing nothing — and many are followed by overshoot, where people end up heavier than they started:

• Very-low-calorie diets. Fast loss, fast metabolic adaptation, common complete regain within a year of stopping.
• Keto, OMAD, juice cleanses, “detoxes.” Restrictive protocols whose effects are mostly mediated by eating less and the novelty of a new system. When the novelty wears off — usually by month three to six — adherence collapses on the same curve as any other restrictive diet.
• Cheat days. Often net zero weekly deficit, while training the body to associate one day a week with abandoning the framework. That’s the opposite of what builds a habit.
• Goals tied to a date. “Lose 20 lbs by the wedding.” Intense short-term effort followed by total disengagement once the date passes. The structure that held the behavior in place dissolves with the deadline. There’s nothing underneath.

What I do instead

I work with people on the boring version of this. Not because I don’t care about results — because the boring version is the only one with results that hold past year two.

That looks like:

• A nutrition pattern built around what you actually eat, whether you enjoy it, and whether you can hold it on a Wednesday in February.
• Strength training as the lifestyle anchor — because muscle does most of the metabolic work behind the scenes.
• A weekly call where you can be honest about what didn’t happen, and we adjust without anyone losing their nerve.
• No goal with an end date. The goal is a way of living that still works in five years.

One last thing

The DPP wasn’t magic. The intervention was modest. What made it work was that someone — a coach with a curriculum and ongoing contact — was there, paying attention, while people did the boring thing for long enough that it stopped being effortful.

That’s the part that’s missing from almost every plan, app, and AI you’ve tried. Not the plan. The person.