does diet coke actually mess with fat loss.

Published October 5, 2026 nutrition science 16 min read

a two-year randomised trial put diet drinks against water in 493 people trying to lose weight, + the diet-drink group lost slightly more. here is what the trial measured, who paid for it, who dropped out, + the honest answer to whether the can in your hand is costing u anything.

why everyone is suddenly arguing about diet coke again

last week a university of liverpool trial went through the whole cycle in about four days. sciencedaily, yahoo, studyfinds, the uk tabloids. the sentence was the same everywhere: diet drinks work as well as water for weight loss. + the replies underneath were the same everywhere too. one half said finally, the other half said of course the soda industry paid for it.

both halves are half right, which is why i want to do this one slowly.

i also have skin in this. the fighter program has a two-week primer phase, + artificial sweeteners are on the list of things it cuts. the babe-science line in the recipe section says it out loud: the fewer artificial sweeteners u eat, the sweeter food starts tasting. i am not retracting that tonight. i am going to tell u exactly which part of it the trials back + which part is me doing a thing i like.

the wellness side has a louder version. matt at endsickness, who i rate higher than almost anyone on the mould + histamine lane, puts it this way: sugar-free usually means we swapped sugar for lab chemicals, + he lists aspartame, sucralose + acesulfame as things that confuse blood sugar signalling, irritate the gut + worsen cravings. eva's version is two words: skip the soda. i want to name those pitches properly before i take them apart, because the named version is always stronger than the internet says, + parts of it survive.

what the trial actually did

here is the design, plainly, because the design is the whole argument.

493 adults with a BMI of 27 to 35 who already drank cold drinks regularly were randomised to one of two arms: at least two 330 ml servings a day of water, or the same of a diet drink, chosen from about 20 branded options, each under 2.1 kcal per 100 ml. the water arm was also told to stop all sweeteners, including the ones u stir into coffee. everyone then joined the same group behavioural weight-management program: weekly meetings for 12 weeks, then monthly for 40 weeks, then a voluntary unassisted year with one final weigh-in at week 104.

the question was an equivalence question, which is a specific kind. a normal trial asks is A better than B. an equivalence trial asks is A as good as B, within a margin the researchers set in advance. if the two arms land inside that margin, they are declared equivalent. that is the claim being made here, + it is a narrower claim than the headline.

two things the coverage skipped. first, at the one-year mark the arms were not equivalent: the diet-drink group was significantly ahead by 1.4 kg, + the authors wrote in the same paragraph that 1.4 kg is not clinically significant. second, by two years, with the weekly meetings gone, both groups had drifted back up by about two and a half kilos + the gap closed. the thing that moved the scale was the program. the drink was never the lever either way.

who paid for it, + what that does + does not mean

the trial was funded by the american beverage association. that is the trade body for coca-cola, pepsi + the rest. the papers state it in the funding line, + they state that the sponsor was not involved in design, conduct, analysis or interpretation, + that a sponsor representative was given the chance to review the manuscript.

i am not going to pretend that does not matter. industry-funded nutrition trials are more likely to reach industry-friendly conclusions, + the reason is rarely fraud. it is which questions get funded at all. nobody at the beverage association is paying to find out whether water beats diet coke for your liver enzymes over ten years. they paid for a question where the drink could win or draw.

here is the part that keeps me honest in the other direction. the 2015 bristol meta-analysis that pooled every sweetener trial had co-authors from danone, unilever + an industry science body, and the whole point of pooling is that single-sponsor effects wash out across studies with different sponsors. across 129 short-term comparisons, a sweetened drink before a meal changed what people ate at that meal by −2 kcal versus water, with a confidence interval of −30 to +26. across the longer trials, sweeteners in place of water came out 1.24 kg lighter. that is the direction every trial points, whoever paid. the sponsor is a reason to read the paper harder. it is not a result.

the dropout, which is the real weakness

this is where i would push back if i were reviewing the paper, + the authors push back on themselves.

of the 493 randomised, 383 made it to week 12. 262 made it to week 52, which is 53 percent. 220 made it to week 104. so the headline number is a weight from fewer than half the people who started, + the people who stay in a two-year weight-loss trial are not a random sample of the people who join one. they are the ones it was working for. the authors re-ran the analysis with the missing people statistically filled in [two separate methods] + got the same answer, which helps. it does not make 220 into 493.

it was also one site, in england, with no ethnicity data, run partly through covid, + people were paid £300 for finishing the first year + £100 for the second. the authors list all of it. i am listing it because a result this convenient deserves its limitations printed next to it, not three paragraphs down.

is this one trial, or is this the pattern

it is the pattern, + that is the part that should move u more than any single study.

in 2016 a colorado group ran the same design for one year with 303 people drinking 710 ml a day of water or diet drinks inside a weight-loss program. water kept off 2.45 kg. diet drinks kept off 6.21 kg, p < 0.001. the honest caveat on that one: everyone in it was already a regular diet-drink user, so the water arm was being asked to give something up, which is its own stress.

in 2012 the CHOICE trial took 318 people who drank at least 200 kcal a day of sugared drinks + swapped them to water, diet drinks, or a control group that made whatever diet change it liked. six months later: diet drinks −2.5 percent of body weight, water −2.0 percent, control −1.8 percent, no significant differences. the water group did get one thing the others did not: a significant drop in fasting glucose versus control, p = 0.019.

then the pooled view. a 2022 network meta-analysis in JAMA network open took 17 randomised trials + 1,733 adults + asked three questions at once. swap a sugared drink for a diet one: −1.06 kg, less body fat, less liver fat. swap a sugared drink for water: no measurable change in any outcome, from only 3 trials. swap a diet drink for water: no weight difference, a slightly worse HbA1c with the diet drink [+0.21 percent] + a slightly better systolic blood pressure [−2.6 mmHg]. certainty was moderate for the first swap + low for the other two.

so why does the WHO say the opposite

because it is reading a different kind of study, + this is the most useful thing on this page.

in 2023 the world health organization issued a conditional recommendation against using sweeteners for weight control, saying they do not confer any long-term benefit in reducing body fat. conditional is their own grading: it means the evidence was low certainty + the call could go differently in different places. the review behind it leaned on cohort studies, the kind that follow people for years + record what they drank + what happened to them.

in cohorts, diet-drink users do worse. they gain more weight, they get more diabetes. in randomised trials, where a coin decides who drinks what, they do not. that pattern has a name: reverse causation. people switch to diet drinks because they are already gaining, already pre-diabetic, already trying. the drink is a marker of the problem, not the cause of it. the bristol review said it in one line: twelve cohort studies, inconsistent associations, −0.002 BMI units per year, with reverse causation the likely explanation.

a 2019 review in the BMJ that did it the cochrane way came to the same shrug: across 56 studies, most outcomes showed no difference between sweetener users + non-users, the few trials were small + short, + confidence in all of it was limited. so the honest sentence is not sweeteners are safe. it is: nobody has ever shown they cost u fat loss, + the studies that look like they do are the ones where the people chose their own drink.

the part of the wellness argument that survives

now the other direction, because i do not want u leaving here thinking the can is water.

in 2022 a weizmann institute trial gave 120 healthy adults two weeks of saccharin, sucralose, aspartame or stevia sachets, at doses under the acceptable daily intake, against a glucose sachet + a no-sachet control. all four sweeteners changed the stool + oral microbiome + the blood metabolites. but only two moved the thing that matters: saccharin + sucralose significantly impaired glucose responses. aspartame + stevia did not. then they transplanted the human microbiomes into germ-free mice, + the mice reproduced their donors' glucose responses. that is a real finding, it is human, + it is sweetener-specific.

it lines up with a 12-week purdue trial from 2019: 154 people randomised to 1.25 to 1.75 litres a day of a drink sweetened with sugar, aspartame, saccharin, sucralose or stevia. sugar + saccharin both gained weight [+1.85 kg + +1.18 kg]. aspartame + stevia held. sucralose went slightly negative. glucose tolerance was unaffected by any of them at 12 weeks. the authors' conclusion is the one i want u to keep: sweeteners are different molecules + should be judged separately. "artificial sweeteners" as one word is where the argument goes wrong on both sides.

+ in the two-year trial itself, water quietly won two small things: a significantly larger HbA1c reduction + a larger diastolic blood pressure drop in the water arm at week 104. hunger ratings did not differ. total sweetener intake fell in the water group + stayed flat in the diet group, which tells u something true: the diet drink keeps the sweet habit alive, even when it is not keeping the weight on.

read your own label. brands change sweeteners by country + by year. the trial findings are per molecule, not per logo.

what actually moved the scale in that trial

both groups lost six to seven kilos in year one. the drink did not do that. the weekly group meeting did. a nutritionist, a diary, a weigh-in, twelve weeks of weekly contact + then monthly. when the contact stopped in year two, both arms regained about 2.4 to 2.7 kg + landed within a kilo of each other.

that is the finding i would screenshot. the variable that predicted the weight was whether someone was still showing up. the drink in their hand was noise around it. it is the same shape as every nutrition argument i have ever watched people lose months to: the thing u are arguing about is a rounding error on the thing u are not doing.

+ it is where matt + eva are right in a way the trials cannot capture. a diet coke is not a sign of a problem. six diet cokes a day in place of meals is. the trials tested two cans. nobody has tested the 2-litre bottle on the desk, + i am not going to pretend they have.

what i actually do, + what i will not tell u it does

the fighter primer phase still cuts artificial sweeteners for two weeks. i am keeping it. here is the honest version of why.

the primer is a reset: two weeks off sweeteners, starches, dairy + alcohol so u can feel what your baseline is + add things back one at a time. that is a practice, + it is mine, + it costs nothing. what i will not tell u is that the sweetener part is why the scale moves. the trials say it is not. the taste-recalibration claim, that food tastes sweeter after two weeks off, is something i notice in my own mouth + something the 2019 BMJ review could not find a trial for either way. so: i do it. i like it. the record on that specific claim is thin. the ritual stays + i am not taking it off u. i just will not tell u it is the medicine.

- MoMo

questions i get at 1am

does diet coke stop fat loss

No randomised trial has ever found that. In the two-year University of Liverpool trial, people drinking at least two diet drinks a day lost 4.8 kg against 3.7 kg for water, a difference that was not significant. At one year the diet-drink group was ahead by 1.4 kg. A 2016 trial of 303 people and the 2012 CHOICE trial of 318 people point the same way. The cohort studies that show diet-drink users gaining weight are the ones where people chose their own drink, and people who are already gaining are the ones who switch. The drink is a marker, not a cause.

is water better than diet soda for losing weight

Not for weight, on the evidence. Across every randomised trial, water never beat diet drinks on the scale. Water did win two small secondary outcomes in the two-year trial: a larger HbA1c reduction and a larger diastolic blood pressure drop. A 2022 meta-analysis found replacing a sugared drink with a diet one cut 1.06 kg and reduced liver fat, while replacing it with water showed no measurable change in any outcome from only three trials. Water is a good choice. It is not a lever.

do artificial sweeteners raise blood sugar

Depends which one, which is the point most arguments miss. In a 2022 trial of 120 healthy adults, two weeks of saccharin or sucralose significantly impaired glucose responses, while aspartame and stevia did not, even though all four changed the gut microbiome. In a separate 12-week trial of 154 people, glucose tolerance was unaffected by any sweetener, but saccharin and sugar both caused weight gain. Diet Coke is aspartame. Most sugar-free energy drinks are sucralose. Read the label and judge the molecule, not the category.

was the diet drink study funded by the soda industry

Yes. Both the one-year and two-year papers state the trial was funded by the American Beverage Association, that the sponsor was not involved in design, conduct, analysis or interpretation, and that a sponsor representative could review the manuscript. That is a reason to read it harder, and the real weakness is that only 220 of 493 people finished two years. But the result matches the 2016 Colorado trial, the 2012 CHOICE trial and the pooled meta-analyses, which had different sponsors and the same direction.

+ if the number on the scale is running your day, that is worth saying out loud to someone. in the US, call or text 988. free, confidential, 24/7.

keep reading

• why am i not losing weight if i'm eating less. — the four things that are actually stalling it, none of which are in a can
• why am i starving after the gym. — the real craving mechanism, with the compensation numbers
• why did i gain weight overnight. — what a litre of anything does to the scale by morning
• do i actually need 50 grams of fiber. — the other viral nutrition claim, taken apart the same way
• does creatine work if i don't lift. — last week's headline trial, read all the way down

Published October 5, 2026 · sources are the published trials, meta-analyses + guideline named in each line; weights, p-values + confidence intervals are quoted as printed in the papers. Funding statements are quoted from the papers' own funding sections.

this is education, not medical advice. nothing here diagnoses or treats a condition. if u have diabetes, PKU [aspartame carries a phenylalanine warning for a reason], or take medication that interacts with what u eat, your clinician owns that call, not a blog.