What diet changes can achieve
Average weight loss from sustained dietary changes in clinical trials, compared with prescription medication.
The bottom line on diet and weight loss
All dietary approaches produce similar weight loss when calories are matched and people keep going. The critical question is not "which diet works best?" but "which approach can you maintain long-term?" High protein intake is the one consistent difference-maker. It protects muscle whichever eating pattern you follow. If you are on GLP-1 medication, diet still matters. What you eat shapes how much of the loss is fat rather than muscle, even when the drug handles appetite.
Caloric deficit
An energy deficit means taking in fewer calories than the body uses. It is the fundamental mechanism behind all dietary weight loss, and it is not contested in nutrition science. A deficit of approximately 500 kcal/day produces about 0.5 kg of fat loss per week in controlled conditions.
The evidence behind these numbers
A 2024 review pooled 47 randomised trials with 3,363 participants, comparing various ways of eating less. All of them produced real weight loss when people stuck with it: eating less every day, alternate day fasting, and eating within a set window.
As weight is lost, metabolic rate decreases and hunger hormones increase. The deficit needed to keep losing weight gets harder to hold over time. Very large deficits, over 1,000 kcal/day, cost more muscle and are harder to sustain.
High protein diet
Of all the ways to shift what is on the plate, higher protein has the most consistent evidence. It protects muscle during weight loss, and it produces slightly greater fat loss than lower-protein diets at the same calorie level.
The evidence behind these numbers, and good protein sources
A review pooling 24 randomised trials compared high-protein, calorie-restricted diets with standard-protein ones. The high-protein diets produced greater weight loss (−0.79 kg), greater fat loss (−0.87 kg), more muscle kept (+0.43 kg), and more calories burned at rest (+142 kcal/day). The recommended target is 1.2–1.6 g protein per kg body weight per day during active weight loss.
This is especially important for people on GLP-1 medications, which can cause rapid weight loss. Higher protein helps ensure more of the weight lost is fat rather than muscle. Good sources include chicken, fish, Greek yoghurt, eggs, lentils, and tofu. If food alone doesn't get you there, see the trial evidence on protein supplements. Some people also add a fibre supplement alongside meals — see the trial evidence on psyllium husk.
Low-carb vs low-fat
When the two most popular dietary approaches are compared directly in randomised trials, the differences are smaller than most people expect.
The most rigorous head-to-head comparison found no significant difference between healthy low-carb and healthy low-fat diets. Both produced approximately 5–6 kg of weight loss at 12 months. Neither genetic markers nor insulin levels predicted which diet would work better for a given individual.
A broader review across more trials
A broader review of 33 randomised trials (3,939 participants) found a modest 1.33 kg advantage for low-carb diets. Low-carb shows better improvements in triglycerides and HDL cholesterol. Low-fat shows a bigger drop in LDL cholesterol. The practical takeaway: both work, and the best choice depends on which approach you find easier to sustain.
Intermittent fasting
In clinical trials, intermittent fasting and time-restricted eating produce similar weight loss to eating less every day. The main advantage is that some people find them easier to stick to than daily calorie counting.
The evidence behind these approaches
A 2024 review of 47 randomised trials ranked alternate day fasting as producing the greatest weight loss, ahead of daily calorie restriction and time-restricted eating (16:8). However, the differences are modest. Most of the weight loss comes from eating fewer calories in total, not from any special effect on metabolism.
If you are on GLP-1 medication, check with your healthcare provider before adding fasting. The drug's appetite suppression already creates a significant calorie gap. Adding fasting on top can make it harder to get enough protein and nutrients.
Ultra-processed food
A landmark 2019 NIH inpatient trial gave the first direct experimental evidence that ultra-processed food causes people to eat more. That held even with calories, protein, fat, carbohydrate, sugar, sodium and fibre matched.
Participants ate as much as they wanted from either an ultra-processed or an unprocessed diet, for two weeks each. The meals were matched for available calories and for protein, fat and carbohydrate. Even so, participants ate 508 more calories per day on the ultra-processed diet. The mechanism appears to be eating speed: ultra-processed food is eaten faster, and fullness signals cannot keep up.
What counts as "ultra-processed"
Not all processed food is problematic. The NOVA classification system sorts food into three groups: minimally processed (fresh, frozen), processed (canned, preserved), and ultra-processed (industrially manufactured with additives). The evidence specifically concerns ultra-processed products. Cutting them back and eating whole foods instead naturally reduces calorie intake in most people.
If you're also taking medication
Diet doesn't stop mattering when you start a weight loss drug. It shifts. GLP-1 medications reduce appetite, which makes it easier to eat less. But what you eat still matters for keeping muscle, for energy and for long-term health. Put protein first (1.2–1.6 g/kg/day), eat whole foods where possible, and think about the quality of the food rather than counting calories. The drug handles the calorie gap.
A note on diet trials
Diet research is hard to conduct. People can't be blinded to what they eat, so they always know which diet they are on. Long-term results depend more on whether people stick with it than on which diet they follow. Take all "best diet" claims with scepticism. The evidence consistently points to sustainability as the critical variable.
Written and maintained by one person · Built with agentic AI tools · Every number checked against the cited trial · Not medical advice
Next step most people take
Built with agentic AI tools and not a substitute for medical advice
Last reviewed: