Carbohydrates, calories, appetite, and body weight.

The Optimal Diet, Atkins, South Beach, Paleo, Zone… all have one thing in common: some degree of carbohydrate restriction.

Low, lower, lowest: does it matter?

There are 4 relatively large, randomized ‘diet-induced weight loss’ studies that all reported fairly comprehensive food intake and body composition data. The studies ranged in duration from 24 weeks to one year and included anywhere between 50 and ~300 overweight and obese participants.

In general, participants assigned to the low fat intervention were advised to restrict calories and fat whereas those assigned to low carb were told they could eat as much as they wanted as long as it wasn’t carbs.

Your mileage may vary – but these studies cover a large number of subjects from a wide range of backgrounds, suggesting the results might be applicable across the board.  Conclusion?  the amount of body fat lost was much more strongly associated with the reduction in carbohydrates than calories.  The only modestly surprising aspect was the magnitude… (see the figures below).

The four studies, in chronological order:

Brehm 2003: over the course of 6 months, those who consumed an average of 163 grams of carbohydrate per day lost 8.6 pounds of body weight while those who consumed 97 grams lost 18.7 pounds.

McAuley 2005: 24 weeks; those who ate 171 grams lost 10.3 pounds, while those who ate 133 grams lost 15.2 pounds, and those who ate 107 grams lost 15.6 pounds.

Maki 2007: 36 weeks; those who ate 186 grams lost 5.7 pounds, those who ate 131 grams lost 9.9 pounds.

Gardner 2007: 1 year – those who ate 138 grams lost 10.3 pounds, 181 grams lost 3.5 pounds, 195 grams lost 4.8 pounds, and 197 grams lost 5.7 pounds.



Note in chart above; obese participants assigned to low[ish] carbohydrate diets (red) spontaneously reduce calorie intake, almost to the same degree as those assigned to calorie-restricted, low fat diets.

Plotting all of the data together yields a rather poor relationship between calorie intake and weight loss:

calories vs BW


And the picture doesn’t change much if body weight is replaced by fat mass:

calories vs fat loss

However, if calories are replaced by carbohydrates, the association is much stronger:

carbs vs BW

Same holds true for fat mass:

carbs vs fat loss

Part II

“those assigned to low carb were told they could eat as much as they wanted as long as it wasn’t carbs.”

1. Carbohydrate restriction is almost always accompanied by a spontaneous reduction in appetite; “ad lib” low carb works very well for obese patients.  Once a reduced body weight is achieved, “ad lib” may not work so well as evidenced by near universal weight regain (regardless of diet).

In some of the studies, those assigned to low carb ended up eating slightly more than those on low fat (see graph above).  Interestingly, however, those assigned to low carb still lost more weight and body fat.

2.  It’s not magical; the [notorious] Ebbeling study showed low carb = better preservation of energy expenditure in weight-reduced people.

The Laws of Energy Balance.  They work.

calories proper

Be Sociable, Share!
  • charles grashow

    There’s also this 24 week study

    Long-term effects of a ketogenic diet in obese patients

    To determine the effects of a 24-week ketogenic diet (consisting of 30 g carbohydrate, 1 g/kg body weight protein, 20% saturated fat, and 80% polyunsaturated and monounsaturated fat) in obese patients.


    The mean initial weight of the subjects was 101.03±2.33 kg. The weight decreased significantly during all stages of the treatment period. The body weights at the eighth, 16th and 24th week were 91.10±2.76 kg, 89.39±3.4 kg and 86.67±3.70 kg, respectively. Similar to the loss in body weight, a significant decrease was observed in the BMI of the patients following the administration of the ketogenic diet. The initial BMI, and the BMI after the eighth, 16th and 24th week were 37.77±0.79 kg/m2, 33.90±0.83 kg/m2, 33.24±1.00 kg/m2 and 32.06±1.13 kg/m2, respectively.

    • William Lagakos

      Hi Charles,
      Thanks for the link. They didn’t report detailed diet or body composition data, but seems in line with the other studies… although the participants lost significantly more weight (14%).

  • Spittin’chips

    I caught part of a TV show the other night where they were doing an experiment on a group of willing participants. It seemed to be testing the hypothesis that if you starve and torture obese people, they cry a bit, then thank you for it and lose shed-loads of ‘weight’.

    Can’t remember what it was called but it looked very scientific and totally debunks your piffling little studies here…Exploiting Desperados? no, that wasn’t it. Biggest Farce? no….doesn’t matter. The torturers looked very lean and that’s all that matters.

    • wenchypoo

      The Biggest Loser? Others might call this summer camp. :)

      • Spittin’chips

        That’s the one! :) I try not to leave small, heavy items in the lounge room, because if I happen to catch any part of that program, I get an overwhelming urge to throw something at the TV. It doesn’t help that my gym has a life-size cut-out of the most annoying torturer just inside their front door. Training angry can lead to injuries.

        • wenchypoo

          unsubscribe me.

          • Wenchypoo

            That was for the comment replies ending up in my inbox. I come here to read them personally.

  • RIchard Feinman

    It is good that you are looking at meaningful variables rather than “diets” but for dependent variable (which is usually on the Y axis) you might look at triglycerides which are more reliably related to metabolism. My re-plot of Gardner’s data is on Pinterest at

    • William Lagakos

      Wow, those triglycerides really do fall into place! R-squared = 0.875…

  • Markymark

    I agree that low carb is certianly the way to go if you want to lose weight. The problem is what to do when you have reached your desired weight and your goals change. I’m paleo-primal and happy with the weight but I’m now trying to maximise my testosterone naturally. To max testosterone most people recommend high carb, high fat and moderate protein.

    • William Lagakos

      Hi Mark,
      Regarding carbs and testosterone (in men), I came across a few studies which suggest otherwise… wrote about them here:

  • CynicalEng

    How many of the studies combined men and women ? I find this always reduces the statistical significance by increasing the variance within the group. Keeping them separate gives clearer relationships between variables like calories eaten and weight loss as the men’s and women’s metabolic rates will cluster in two largely separate groups with some modest overlap.

    Access to the raw data of all trials would greatly help tease out this kind of issue.

    • William Lagakos

      Brehm: women
      McCauley: women
      Maki: 2/3 women
      Gardner: women

      “Access to the raw data” …in my experience, researchers will frequently share data and answer questions (within reason).

  • Chris

    It’s very clear that low carb = bodyweight loss and most importantly body fat loss. However, I’m more into the research of strength training and anaerobic energy production under ketogenic diets.

    How familiar are you with that Bill?

    • William Lagakos

      Pretty familiar. I think it all comes down to adequate ketoadaptation. The rate of energy production from anaerobic glycolysis shouldn’t be too far ahead of that from fatty acid & ketone oxidation if mitochondrial function and number is increased.

      • johnnyv

        Also remember phosphocreatine + ADP -> creatine + ATP for all you 1RM needs! Require adequate rest periods though.

        • William Lagakos


      • Chris

        I’ve personally noticed an increase in my performance in weight lifting as well as endurance exercise and sprinting once I got keto-adapted. So, I’m not sure what to attribute this to given all the hype of “you cannot do explosive exercise on low carb”.

        • William Lagakos

          Given the findings of Paoli and of course Phinney et al., I think there’s some mythology mixed in with the dogma here. High performance can be achieved with a variety of diets; don’t always need to carb-load.

          • Danny J Albers

            However carb loading is certainly a much simpler path, carbs are readily available in rapidly digesting forms today thanks to a plethora of pre and post workout concoctions.

            Personally I find the biggest limiter on the power of a workout is your brain and it will, when you are really killing it, find all kinds of excuses to listen to your body and quit or cut back on intensity.

            The PFP crowd are great examples of being explosive while keto and posting pretty impressive numbers in the weight room and maintaining a pretty potent CF schedule on top.

          • William Lagakos

            …still waiting for Steve Phinney (or more likely Jeff Volek) to publish some real data on the impact of ketogenic dieting on anaerobic/strength/resistance/high intensity/etc….

            until then:

          • William Lagakos

            also, I agree, motivation and psychology are big factors.

            Analogous to long-distance runners… wrt marathon performance, I think mental preparation outweighs “carb-loading” or other nutrition strategies (at least to a degree).

  • sylvie

    Have you come across any compelling studies to suggest that carb-restriction for non-obesogenic women can have a negative impact? Paleo for Women and Gnolls both suggest this, and there are anecdotal stories (as there are for everything), but I am wondering a) if you find possible explanations satisfactory; and b) whether hormonal contraceptive would interfere. Thanks!

    • William Lagakos

      Hi Sylvie,
      I haven’t come across any studies showing a negative affect; most show the opposite. Do you have a link for Stanton’s article?

      Regarding contraception, I don’t know! Did you have a specific concern? (I’d be happy to look into it)

      • sylvie

        The link for the Paleo women’s site was as follows:

        But Stanton didn’t actually give a link, he simply mentioned it in the post. When I asked him to elaborate on the comment, he said the following: Note that I didn’t say IF isn’t a good idea for reproductive-aged women! I said that I’ve observed that a substantial number of them don’t seem to do well on the classic 16/8 regimen. I don’t know enough endocrinology to speculate as to why, but I offer the following observations:

        1. A good metric for whether IF is a good idea for you is whether you can do it without feeling like you’re starving.

        2. You don’t have to IF every day.

        3. Therefore, if you’re really hungry for breakfast one day, eat it!

        4. Skipping breakfast isn’t the only possible plan. If scheduling permits, I know people who have success with a big, hearty late breakfast once they finally get hungry in the morning (~10 AM), skipping lunch, and then eating dinner normally.

        5. A spoonful of coconut oil can sometimes help get you through a rough spot (i.e. you fasted and maybe shouldn’t have, but can’t get to real food for another few hours).

        I think part of the problem might be that women are so used to restrictive diets that always feeling hungry is shrugged off as the price of not being fat. It’s easy to starve yourself into a low thyroid state with that mindset.



        • sylvie

          PS The link above is not a generic link, and the author lists a number of studies on mice and humans that ostensibly support her view that IF is bad for women.

          • William Lagakos

            Thanks for that link! I’ve read the human studies she cited (even blogged about them), but will re-read; this time focusing on gender effects.

            I’d expect IF (or any type of fasting) to impact ovulatory function, but *causing* weight gain seems odd. Will check back.

        • Thomas Hemming Larsen

          You don’t need to be a woman to feel starving when intermittently fasting. I’ve tried to fast as little as 12h and felt very hungry despite that I’m eating a ketogenic diet and have been doing so for more than one year.

    • J. Stanton

      sylvie: The articles you quote, including my response to your comment, are about intermittent fasting, not carb restriction — which is a different issue entirely.


      • sylvie

        True, I’ve been lumping them unintentionally into the same category because people who IF–at least, the anecdotal accounts I’ve read–often seem to consume lower calories, and lower carbs. Moreover, many find IF more manageable when they are keto/low carb due to the suppression of appetite.

  • RIchard Feinman

    Good post but I feel that you are implicitly buying into the assumption of that there is something inherently superior to an RCT, an idea invented by the medical nutrition establishment. You know, the friendly folks who brought you the obesity epidemic. Part of the “levels of evidence” idea that is unknown in any physical science and basically consistent with the idea “physicians prefer a large study that is bad to a small one that is good.” The idea that one type of experiment answers all scientific experiments is dumb. RCT’s are good for some things. The studies you cite are poorly controlled because they are long term and mix in adherence, Smallers studies have better control and if you attribute adherence to the diet itself without proof that it wasn’t poor followup, inadequate instruction to begin with scare stories from American Heart Association or whatever, well, that’s dumb too. And the tradition tied ins some way to the poor control of the RCT is reporting group statistics — nobody loses an average amount of weight. If you want to see what you can do with a diet, look at small well designed studies. I recommend you present some of the studies from Volek’s lab and I will follow these comments up on my own blog.

    • William Lagakos

      yes sir!

      (regarding studies from Volek’s lab)

  • Seth Roberts

    shouldn’t you have plotted rate of weight loss, not total weight loss? since the studies last quite different lengths of time.

    • William Lagakos

      Sure, and that would’ve certainly added to the picture. However, those data weren’t reported – we usually only get baseline and one or two follow-up time points.

      • Seth Roberts

        Those data weren’t reported? You get the rate of weight loss by dividing total weight loss by length of study. I seem to be misunderstanding something.

        • William Lagakos

          Oh! It was me who was misunderstanding.

          This would make the longer studies appear less favorable bc weight loss is usually quite rapid at first, then tapers off… ideally, I should try to find a few studies of the same duration.