Ketogenic diets and weight loss beyond the first few months

Keto can reduce weight, but the long-term advantage is modest. Read the evidence on body fat, muscle, cholesterol and supervised low-calorie diets.

日本語で読む（Japanese version）

Is cutting out rice the same intervention as cutting your entire daily food intake to a very low level? Both can appear under the label “keto.” Yet the research includes diets allowing no more than 50 g of carbohydrate a day and clinical programs providing approximately 500–800 kcal in total. The names overlap; the meal plans and the involvement of health professionals differ substantially.[1][6] Before comparing dramatic weight-loss stories, it helps to establish which kind of diet each story describes.

A ketogenic diet can be an option for weight loss. In a meta-analysis of long-term comparisons in adults, it produced more weight loss than a low-fat diet. The average additional loss, however, was 0.91 kg.[1] That is a different proposition from a large advantage that keeps widening as the months pass. Adding the word “ketones” to a meal plan does not promise what the scale will show next year.

For someone choosing meals around work, family dinners and shopping in Tokyo, the useful questions concern the conditions behind the results. How tightly was carbohydrate restricted? Was total energy also restricted? Who took part, and how long were they followed? This article examines strict carbohydrate restriction through those questions, keeping weight, body composition, exercise performance and blood lipids separate. A diet deserves to be judged by the outcomes measured, rather than by the confidence of its name.

The ketogenic label does not describe a single meal plan

Bueno and colleagues defined the very-low-carbohydrate ketogenic diet in their long-term meta-analysis as no more than 50 g of carbohydrate per day. The comparison was a low-fat diet that restricted energy and supplied less than 30% of total energy from fat. The researchers collected randomized trials in adults with at least 12 months of follow-up.[1] In a randomized trial, allocation to the diet groups is arranged so that participants’ preferences alone do not determine which group they join.

This comparison matters because both groups were following a dietary intervention. It did not compare a ketogenic plan with a life in which nothing about food changed. When a result says that one group lost more weight, the relevant question is “more than what?” An energy-restricted low-fat diet is an active comparison. Its presence changes the meaning of the advantage reported for the ketogenic group.[1]

Muscogiuri and colleagues examined a different intervention: a very-low-calorie ketogenic diet. Alongside less than 50 g of carbohydrate per day, it provided approximately 500–800 kcal daily and 15–30 g of fat. Protein was set in relation to ideal body weight.[6] These figures describe a clinical program. They are not a set of targets to copy at home after reading an article. The review made individualization, consideration of contraindications and supervision by a health professional part of the approach.[6]

Removing a staple food and replacing it with other foods is therefore a different decision from entering a professionally designed program that also sharply restricts total energy. A headline about substantial weight loss can conceal that distinction. Without the energy prescription and the rest of the diet, it is impossible to tell whether the result describes the kind of change you are considering.[1][6]

The boundary between “low carbohydrate” and “ketogenic” also varies across research. Leung and colleagues’ 2025 meta-analysis examined ketogenic and low-carbohydrate diets together. Its overall analysis included a carbohydrate limit of 100 g per day, while the group consuming no more than 50 g per day was examined separately.[2] Results found under a broad low-carbohydrate search should not automatically be assigned to a stricter ketogenic diet.

Nor do the studies establish that ketones independently account for every kilogram lost. The comparisons assessed assignment to diets and subsequent changes in outcomes. Even when prescribed energy was comparable, actual energy and protein intake remained uncertain.[4] A meal plan can be named precisely while its execution remains less certain. The dietary label is a starting point for understanding the intervention; it cannot replace knowing what participants actually ate.

The long-term advantage exists, but its size matters

Bueno and colleagues’ 2013 meta-analysis included 13 trials. The body-weight analysis covered 1,415 participants followed for at least 12 months. Compared with the low-fat groups, those assigned to ketogenic diets lost an additional 0.91 kg on average. The 95% confidence interval was −1.65 to −0.17 kg, and the difference was statistically significant.[1] For this comparison, the interval shows the range of uncertainty around the additional weight loss estimated for the ketogenic group.

The 0.91 kg figure is not the total amount lost by a person following keto. It is the difference between the dietary groups.[1] Confusing those quantities can make the diet look either less effective or more impressive than the comparison supports. The result concerns an additional advantage over another weight-loss diet, rather than the entire change from a participant’s starting weight.

A useful way to read the result is to keep the two questions separate. “Can people lose weight on this diet?” concerns the change within a group. “Does this diet produce more weight loss than another approach?” concerns the gap between groups. The meta-analysis supports a long-term difference in favor of the ketogenic diet, but the size of that difference remains part of the answer.[1]

The authors described the approach as a possible alternative tool against obesity.[1] That supports considering it as an option. It does not establish a large superiority over other diets. If adopting the diet would mean changing shared family meals, restaurant choices and food preparation after a late shift, place those changes alongside the additional average loss the research found. The practical demands of an option matter even when its statistical result is favorable.

At least 12 months of follow-up also gives this finding a different role from a short-term result. It addresses a longer period, but it does not promise that every individual’s weight advantage will keep growing indefinitely.[1] The measured comparison is useful precisely because its duration and magnitude are explicit. Keeping both in view is more informative than treating any significant result as proof of a lasting, expanding lead.

Large losses in very-low-calorie programs are a separate result

The large losses below concern a clinical very-low-calorie intervention, distinct from ordinary carbohydrate restriction. Such programs require individual planning and health-professional supervision, as described in the European review.[6] Castellana and colleagues’ 2019 systematic review and meta-analysis examined very-low-calorie ketogenic diets in people with overweight or obesity. It included 12 studies. Reported weight changes were −10.0 kg in studies with a ketogenic phase lasting up to 4 weeks and −15.6 kg in studies with a phase lasting at least 4 weeks. Subsequent follow-up suggested that weight loss was maintained for up to 2 years.[13]

Those are substantial reported changes. They nevertheless belong to a clinical intervention that also severely restricts energy. They should not be placed beside the 0.91 kg long-term between-diet difference as if the two figures answered the same question.[1][13] One describes reported weight change under a very-low-calorie approach; the other describes an additional average loss compared with a low-fat diet. The reference point is part of the result.

Muscogiuri and colleagues specified protein at 1–1.5 g per kilogram of ideal body weight in the very-low-calorie program.[6] Ideal body weight, rather than current body weight, was the basis for that prescription. Carbohydrate, fat, protein and total energy were designed together. Taking one of those numbers out of its clinical setting loses the conditions under which the intervention was evaluated.

Their review included 15 studies. Compared with other weight-loss interventions of the same duration, the very-low-calorie ketogenic approach produced greater reductions in body weight, fat mass, waist circumference, total cholesterol and triglycerides. Changes in blood glucose, HbA1c and LDL cholesterol, however, were similar to those achieved with the other interventions.[6] HbA1c is a test marker used to assess blood glucose status. A larger change in weight did not mean a larger advantage in every measured outcome.

Castellana and colleagues also reported a discontinuation rate of 7.5%, with no significant difference from low-calorie diets.[13] The comparison concerns discontinuation in the studies reviewed.

The same review reported a waist reduction of 12.6 cm and a triglyceride reduction of 30 mg/dl. Systolic and diastolic blood pressure fell by 8 mmHg and 7 mmHg, respectively.[13] These were reported changes associated with the intervention. They are distinct from numbers describing an additional benefit over the comparison diet. Reading the direction, size and reference point together prevents a collection of favorable figures from becoming a claim the studies did not make.

Comparable energy prescriptions leave some uncertainty

To ask whether carbohydrate restriction adds an advantage beyond the planned energy intake, it helps to compare diets with similar energy prescriptions. Zurawski and colleagues’ 2026 systematic review studied adults with overweight or obesity assigned to ketogenic or higher-carbohydrate diets. Prescribed energy intake, or the targeted energy deficit, was comparable between the groups.[4]

The body-weight meta-analysis included 6 studies and 259 participants. Ketogenic diets were associated with an additional average weight reduction of 1.49 kg, with a 95% confidence interval of −2.41 to −0.58 kg.[4] This suggests that a small additional loss may occur even when planned energy is similar. It is a finding to retain, rather than dismiss because the advantage is limited.

The certainty of the evidence was rated low, owing to risk of bias and imprecision. The evidence was predominantly short term. Actual energy and protein intake were uncertain, and the review did not include a pooled synthesis of body composition. Together, these limitations reduce confidence in the size and durability of the advantage.[4] A numerical estimate can be clear while confidence in what it means remains limited.

When the analysis was restricted to 3 trials that also met the prescribed protein-matching criterion, the direction of the result remained the same. The estimate was imprecise, however, and compatible with no difference.[4] It would be too strong to say that the effect disappeared. It would also be too strong to say that superiority was established after both energy and protein prescriptions were matched.

“Prescribed the same energy” and “actually ate the same energy” describe different things.[4] A plan can specify comparable meals while daily execution varies. A late finish at work, a missed planned dinner or a serving from the family’s meal illustrates why the distinction matters. The review’s uncertainty about achieved intake prevents the observed gap from being attributed confidently to carbohydrate restriction alone.

This review cannot determine next year’s weight or the amount of fat an individual will lose.[4] Its short-term weight comparison addresses a narrower question. Someone considering strict restriction because it seems much more effective than every alternative should first examine both the small estimated advantage and the low certainty attached to it.

Numbers: ketogenic diets and the outcomes they measure

Long-term comparison: at least 12 months, 1,415 participants, and a mean weight difference of −0.91 kg versus low-fat diets; 95% confidence interval −1.65 to −0.17 kg.[1]

Comparable prescribed energy: 6 studies with 259 participants and a mean weight difference of −1.49 kg; predominantly short-term evidence with low certainty.[4]

Adult body composition: across 18 trials, the mean differences were −1.40 kg in fat mass and −0.81 kg in fat-free mass.[11]

Muscle and exercise outcomes: across 33 studies, no significant differences in muscle mass or strength were identified, although fat-free mass decreased.[17]

Blood lipids in normal-weight adults: 3 studies found a mean LDL cholesterol difference of +1.08 mmol/L; 95% confidence interval 0.37 to 1.79 mmol/L.[12]