Meal replacement shakes: what the trials replaced

Two different products share the word shake, and the evidence behind them points different ways. Meal replacement drinks, the kind used inside weight programmes, beat food-based diets of the same calorie count across 22 studies and 1982 adults, with an effect size of 0.261, which is small. Over how long, the pooled review does not say, and the one duration we can give you is from the diabetes trials further down, a median of 24 weeks. The more meals they replaced, the wider the gap got: where the drinks supplied 60 percent or more of daily energy, the effect size rose to 0.545. Before that reads as an instruction, look at how those arms were run. The trials that replaced the most meals were also the most closely supervised ones, and the diabetes trials on this page were run by clinicians from start to finish. If you are managing diabetes or taking medicine for it, building a diet on replacements is a change to make with your clinician rather than on your own, because the trials behind it were run that way. The honest reading of the pattern itself is that what works is having fewer decisions to make about food. Protein powders are the other product, and across twelve trials on 261 young adults doing resistance training, whey and soy did nothing measurable to lean body mass. Bench press rose 8.87 kg and squat 9.60 kg in the same analysis, which usually means practice and neural adaptation rather than new tissue. Almost every number on this page comes from a supervised programme rather than from a tub bought off a shelf. The table below is a third thing again. USDA holds two entries under the word shake, both fast-food milkshakes at 113 and 148 kcal per 100 g, and no trial on this page tested either of them.

The numbers, per 100 g

What a real portion looks like

• Swapping a liquid for a solid of the same kind cut the next meal by about 55 calories, and the result sat exactly on the line of significance. — crossover feeding studies in adults; the food-intake analysis by viscosity covered 191 participants across 9 subgroups (n = 191 in the viscosity analysis) Evidence B source
  Solid compared with liquid food reduced later food intake by 55.5 kcal (95% CI -111.1 to -0.1, p=0.05, I2 0%); semi-solid against liquid showed no difference; higher viscosity gave a non-significant -66.7 kcal (-144.2 to 10.9, p=0.092, I2 84%)

What your body absorbs

• A high-protein shake left the stomach more slowly and yet people ate exactly as much three hours later. — 19 active and 19 less active older adults (n = 38) Evidence B source
  Gastric emptying half time was longer after the high-protein shake, while later energy intake and appetite ratings did not differ in either activity group; the active group reported lower appetite but emptied the stomach faster

What it does to your health

• Diets built around replacement drinks beat ordinary food diets of the same calorie count, and the more meals they replaced the bigger the gap. — 22 studies, 24 interventions, 1,982 adults who were overweight or obese (n = 1,982) Evidence A source
  Meal-replacement low-energy diets beat food-based low-energy diets with Hedges' g 0.261 (95% CI 0.156 to 0.365, I2 21.9%); where replacements supplied 60% or more of daily energy the effect size rose to 0.545 (0.260 to 0.830)

What people believe that the data does not support

• The European obesity body reviewed the newest trials and said replacements came out stronger while intermittent fasting did not. — 56 new studies, of which 33 randomised trials and 21 meta-analyses or systematic reviews of high-quality randomised trials (n = 56 studies) Evidence E source
  The position states that the role of meal replacements is further strengthened by the new evidence, while for intermittent fasting no evidence of significant advantage over continuous energy restriction was found

Who this works differently for

• In people with type 2 diabetes, liquid replacements shifted weight and blood sugar a little and left blood fats alone. — 9 trial comparisons, 961 overweight or obese adults with type 2 diabetes, median follow-up 24 weeks (n = 961) Evidence A source
  Body weight -2.37 kg (95% CI -3.30 to -1.44), HbA1c -0.43% (-0.66 to -0.19), fasting glucose -0.63 mmol/L (-0.99 to -0.27), systolic blood pressure -4.97 mmHg (-7.32 to -2.62), with no effect on blood lipids; overall certainty low to moderate
• Replacing every meal moved blood sugar and weight several times as far as replacing one. — 17 randomised controlled trials, 2,112 participants with type 2 diabetes (n = 2,112) Evidence A source
  Total replacement against partial replacement gave HbA1c -0.72% versus -0.32% (p=.01), fasting glucose -1.45 versus -0.56 mmol/L (p=.02), body weight -6.57 versus -1.58 kg (p<.001) and BMI -2.78 versus -0.37 (p<.001); benefit was similar whether or not participants used insulin, and both forms were well tolerated
• Before diabetes sets in, the same approach moved weight and cholesterol by amounts the authors call small but real. — 10 trials, 1,254 people with pre-diabetes and features of metabolic syndrome (n = 1,254) Evidence A source
  Body weight -1.38 kg (95% CI -1.81 to -0.95), BMI -0.56 kg/m2, LDL cholesterol -0.18 mmol/L, non-HDL cholesterol -0.17 mmol/L, systolic blood pressure -2.22 mmHg; waist circumference and HbA1c changes described as trivial; certainty low to moderate
• Adding two sachets a day to a lifestyle programme took off more weight and more deep belly fat than the programme alone. — adults with prediabetes and overweight or obesity in eastern China; the replacement was a 46 g sachet giving 212 kcal, 9.2 g protein, 19.2 g carbohydrate, 10.2 g fat and 3.5 g fibre, taken as two sachets replacing one meal or one sachet replacing the staple of two meals (n = 12-week randomised trial) Evidence B source
  Weight fell further with replacements (82.2 to 73.6 kg against the lifestyle-only group, interaction p=0.023) as did BMI (p=0.025) and visceral fat area (110.4 to 80.2 cm2, interaction p=0.003); waist, hip and subcutaneous fat showed no significant interaction

What is still unknown

• Whey and soy powders did nothing measurable to lean body mass in trained and untrained young adults. — 12 studies, 261 trained or untrained adults aged 18 to 30 engaged in resistance training (n = 261) Evidence A source
  No significant effect of either whey or soy protein on lean body mass; supplementation raised peak plasma essential amino acids, and whey improved bench press by 8.87 kg (95% CI 5.95 to 11.79) and squat by 9.60 kg (5.61 to 13.60) with low to no heterogeneity

The bottom line

If you are using a shake to lose weight, the setting matters more than the product does. In type 2 diabetes, nine trial comparisons on 961 people gave 2.37 kg of weight and 0.43 percentage points of HbA1c over a median 24 weeks. The reviewers call several of those changes of marginal clinical significance and grade the whole body of evidence low to moderate. Replacing every meal moved things much further than replacing one, 6.57 kg against 1.58 kg, and that comparison is between subgroups of different trials rather than a head-to-head randomisation. The total-replacement arms were also the most closely supervised. A 6.57 kg loss is a medical programme. Before diabetes sets in, ten trials on 1254 people found 1.38 kg and a small drop in LDL on top of a weight loss programme both groups were already following, and the reviewers sort their own findings into small but important and trivial. None of these trials answers the question people actually have, which is what happens after the replacements stop. Two smaller findings are worth holding. Drinking calories instead of chewing them left people eating about 55 kcal more at the next meal, at p exactly 0.05, with one end of the confidence interval at 0.1 kcal, so take the direction and leave the number. And when 38 older adults drank a high-protein shake, it left the stomach more slowly and they ate exactly as much three hours later. Slower gastric emptying is the usual explanation for why shakes fill you up. Here the mechanism moved and the behaviour did not.

Sources

1. shk-dr-02 · meta-analysis of crossover feeding studies
   There was a borderline significant reduction in food intake for studies comparing solid with liquid food of - 55.5 kcal (95% CI - 111.1 to - 0.1 kcal, p = 0.05, I2 = 0%) however no difference in food intake was seen for comparisons of semi-solid and liquid food.
   Sci Rep, 2020 · checked 2026-09-22
2. shk-dr-07 · randomised controlled crossover trial
   Higher protein intake did not affect subsequent energy intake or appetite ratings in both active and less active groups. Gastric emptying half time was longer following the high-protein milkshake compared to the low-protein milkshake.
   Appetite, 2025 · checked 2026-09-22
3. shk-dr-01 · meta-analysis of randomised controlled trials
   The effect size in which MR-based LED was compared with food-based LED for weight loss was small, favoring MR (Hedges' g = 0.261; 95% CI 0.156 to 0.365; I2 = 21.9; 95% CI 0.0 to 53.6).
   J Acad Nutr Diet, 2021 · checked 2026-09-22
4. shk-dr-08 · position statement with systematic review
   Although the role of meal replacements is further strengthened by the new evidence, for intermittent fasting no evidence of significant advantage over and above continuous energy restriction was found.
   Obes Facts, 2023 · checked 2026-09-22
5. shk-dr-03 · meta-analysis of randomised controlled trials
   Liquid meal replacements in weight loss diets lead to modest reductions in body weight, BMI, and systolic blood pressure, and reductions of marginal clinical significance in body fat, waist circumference, HbA1c, fasting glucose, fasting insulin, and diastolic blood pressure.
   Diabetes Care, 2019 · checked 2026-09-22
6. shk-dr-04 · meta-analysis of randomised controlled trials
   In subgroup analyses, total MR showed greater improvement in HbA1c (-0.72% vs -0.32%; P = .01), FBG (-1.45 vs -0.56 mmol/L; P = .02), body weight (-6.57 vs -1.58 kg; P < .001), and BMI (-2.78 vs -0.37; P < .001) than partial MR.
   J Clin Endocrinol Metab, 2023 · checked 2026-09-22
7. shk-dr-05 · meta-analysis of randomised controlled trials
   The available evidence suggests that incorporating MRs into a weight loss intervention leads to small important reductions in body weight, BMI, LDL-c, non-HDL-c, and systolic blood pressure, and trivial reductions in waist circumference and HbA1c, beyond that seen with conventional food-based weight loss diets.
   Obes Rev, 2024 · checked 2026-09-22
8. shk-dr-09 · randomised controlled trial
   Lifestyle intervention combined with MRs resulted in greater reductions in body weight and VFA than lifestyle intervention alone in individuals with prediabetes and overweight/obesity.
   J Nutr Metab, 2026 · checked 2026-09-22
9. shk-dr-06 · meta-analysis of randomised controlled trials
   The analysis found no significant effect of either whey or soy protein supplementation on LBM.
   J Diet Suppl, 2026 · checked 2026-09-22