Journal of Personal Science: How Much Salt Should I Eat?

by Greg Pomerantz

The Journal of Personal Science, suggested by Tom and encouraged by Bryan Castañeda, will contain articles about using science to help yourself. This is the first one. It previously appeared on Greg’s blog. If you have written something or plan to write something or are thinking about writing something that might be included, please let me know. — Seth

I spent a few weeks this summer conducting a self-experiment on salt sensitivity and blood pressure. The experiment included a three week phase on a low carb whole foods diet with no added salt, followed by a moderately extreme salt loading phase. This post is a summary of my results.

I learned a lot from the experiment and came out of it with at least one bit of useful information. Will I try to restrict salt in my diet? No, I don’t think salt restriction can work for me. From now on I will ensure that I get sufficient salt on a daily basis.


These are the main points I learned during the experiment, from most to least interesting.

1. Salt restriction caused impaired thermoregulation. In hot weather, my cardiovascular system was not able to sufficiently lower my body temperature. This resulted in an elevated heart rate and hypethermia (up to 101.5 degrees in one instance). This can be dangerous, so be careful if you try this at home.

2. No clinically meaningful change in blood pressure. Systolic pressure was unchanged, though salt loading may have caused a small rise in diastolic pressure. This does not rule out long term negative effects from chronic salt loading (see discussion below), but it does show that, as previously discussed, my kidneys seem to basically work and can regulate my blood pressure through the maintenance of fluid and electrolyte balance in response to changes in my sodium intake.
3. Salt reduction may increase susceptibility to skin infections. Three days into the salt restriction phase, I came down with what was probably a staph infection in my right eyelid. This responded to antibiotics but it came back once I went off them. Since adding back salt I have had no problems with skin infections and no more antibiotics.
4. Possible strength loss. I did not perform well in the gym on my usual strength training program.
5. My taste for salt adapts quickly to restriction and loading. I experienced no cravings even when my sodium intake was too low. I can’t just “listen to my body”. Likewise, while the salt loading phase was difficult for the first two or three days, my taste rapidly adjusted to the added salt.
6. Bodyweight changes. I experienced substantial changes in body fluid levels (e.g. 6 pound weight gain within two hours of the transition from the salt restriction to the salt loading phase).

Conclusion: A low carb paleo diet must include added salt (for me). Can others do without? Perhaps, and some scientists such as Loren Cordain and Tim Noakes (e.g. this podcast episode 18 at 1:03:50) seem to think they can. Skip ahead to read my further musings on this question.

Study Design

The experiment had three phases. First, I did a one week lead-in phase (phase I) where I made no changes to diet or salt consumption. The purpose of phase I was to establish a blood pressure baseline through daily morning measurements (see Measurement Methods below).

This was followed by a three-week sodium restricted phase (phase II) during which I did not add any salt to my food. In addition, during phase II only, I avoided naturally salty foods such as shellfish. My sodium intake during phase II was limited to the sodium in the foods I was eating. Note however that there were one or two restaurant meals per week during this time where I was not able to strictly control for added salt. Sodium consumption on phase II was estimated to be between 800mg and 1000mg per day. phase II was originally scheduled for two weeks, but was extended due to the aforementioned infection and antibiotic use.

Finally, phase III was a salt-loading phase during which I added an additional 5 grams of sodium to my diet, for a total of nearly 6 grams of sodium per day including the sodium naturally occurring in my food. The supplemental salt during phase III consisted of hand harvested French Celtic sea salt (Eden Foods, Inc.) and was measured daily on an AMW-1000 digital scale. Because the Eden French Celtic sea salt is approximately 1/3 sodium by weight according to the label, the 5 grams of supplemental sodium per day was provided by approximately 15 grams of sea salt. Note that different varieties of salt will contain different percentages of sodium by weight. Sea salts vary significantly due to variations in residual water content (not, as commonly assumed, by the presence of other minerals). Please consult the label or a friendly analytical chemist for guidance.

The diet throughout this experiment consisted of meat, fish, eggs, coconut oil, butter, and non-starchy vegetables. In addition, I typically consumed a banana, an ounce (28g) of almonds and a bit of dark chocolate each day. Potassium intake was fairly consistent at around 4 g/day. Table 1 shows a typical day’s macronutrient intake. Given the macronutrient ratio, I believe it is likely that the diet was ketogenic.

Table 1. Approximate daily macronutrient intake.

calories percent (calories)
200 6.6%
Protein 155 620 20.5%
Fat 245 2205 72.9%
Total 3025


Measurement Methods

I measured blood pressure daily first thing each morning while seated, with the cuff of an Omron HEM-711 placed on the left upper arm over the brachial artery. I followed guidelines described by Agena et al (see Chart 2 of the linked paper). Each day’s blood pressure value was determined by averaging the first three measurements taken that morning.

My first measurement of the day was typically higher than the average of the second and third measurements (systolic: +5, diastolic: +4, average over all three phases). This is referred to as the “alarm reaction” and is related to the more commonly known “white coat syndrome”, where the presence of a doctor elicits a stress response and therefore an innacurately high blood pressure reading. My alarm reaction seems to be due to the fact that I get slightly stressed out about seeing what my blood pressure is, even when I measure it myself. Therefore I experience a slight rise in blood pressure while waiting to see the first reading each day. I kept all three readings for this experiment. My “true” normal blood pressure is on average slightly lower than these results which include the first “alarm” reading.


I summarized my qualitative findings in the executive summary above. If you skipped that because you are not an executive, you can go back and read it now. Below are graphs showing my blood pressure and bodyweight during the three phases.

Figure 1: Possible mild elevation in diastolic blood pressure during the salt loading phase. Each point is the average of the three morning blood pressure readings for the day. Red = phase I, green = phase II, blue = phase III. Curves from ggplot2 “geom_smooth()” using default parameters.


Figure 2. No change in systolic blood pressure.


Figure 3: Bodyweight.

Figure 3 shows my daily bodyweight, measured each morning before consumption of any food or fluids. Note that my previous health goal (the 415 deadlift) involved an intentional increase in bodyweight and therefore significant excess calorie consumption. My current diet is lower in calories and Figure 3 therefore should show a long term downward trend in bodyweight.

Salt restriction clearly resulted in a rapid decrease in bodyweight over the first few days of phase II. There appears to be a stabilization towards the end of the salt restricted phase. The salt loading in phase III produced a very large initial weight gain, followed again by stabilization around the same level seen at the end of the salt restriction phase. As salt is primarily stored in bones and extracellular fluids, an increase in salt would be expected to correspond to an increase in extracellular fluid (since the body’s bone mass should change slowly). The bodyweight changes shown in Figure 3 therefore reflect changes in extracellular fluid levels. While salt loading at the levels used in phase III produced a large acute change in body fluids, this was restored to normal over approximately 5 days.

Since my extracellular fluid volume was evidently restored within 5 days, it is not surprising that salt loading had no significant effect on my blood pressure. What is somewhat surprising was that there was no evidence of a temporary increase in blood pressure during the few days in which my extracellular fluid volume was in fact elevated. This suggests that there is an additional regulatory element working to restore blood pressure homeostasis at a shorter time scale than the dominant kidney-fluid mechanism previously discussed on the blog here.

Thanks to Mako Hill for guidance with ggplot2, without which these plots would look less nice.


This experiment demonstrated to me that a low carb paleo diet with no added salt is potentially dangerous for me. Impaired thermoregulation is a big deal and would have been a life-threatening issue if I had to hunt for my food in a hot climate. Not only was my body temperature elevated in warm weather, but my pulse was elevated as well, suggesting my cardiovascular system was unable to restore my body temperature to normal. I’m clearly not salt sensitive, and I do not function well with a low salt diet. However, genetic studies suggest the ancestral human genotype is associated with high levels of salt sensitivity and ability to function with very low sodium intakes. How did humans evolve these traits? And why don’t I seem to have them?

A Faustian Kidney Bargain

Susumo Watanabe has proposed in interesting hypothesis about the evolution of sodium metabolism in hominids. The theory is laid out in a 2002 paper called “Uric Acid, Hominid Evolution, and the Pathogenesis of Salit-Sensitivity,” published in the journal Hypertension. It goes something like this. At some point during the evolution of our common ancestor with gorillas and chimpanzees, a series of mutations inactivated the gene for urate oxidase, an enzyme that breaks down uric acid. As a consequence, we have much higher blood levels of uric acid than other mammals. These mutations seem to have occurred between 24 and 8 million years ago, during the miocene, when our ancestors were believed to be subsisting primarily on fruits and leaves. This diet would have been exceptionally low in sodium. Since there is evidence of multiple independent mutations in this gene in multiple primate lineages, it is thought that mutations deactivating urate oxidase were strongly selected.

In rats, uric acid raises blood pressure acutely, but also causes renal vascular disease via renin/angiotensin systems. This over time makes the rats more salt sensitive. If there is very little salt available, salt sensitivity can be a good thing. Watanabe argues that, where salt is scarce, high uric acid is beneficial (via multiple pathways) for preventing blood pressure from going too low.

In addition to causing kidney disease, high uric acid causes other problems, like gout, and is associated with heart disease. So this looks like an engineering tradeoff with a number of downsides, but some benefits in the context of a miocene diet that was even lower in sodium than the lowest current estimates for paleolithic diets. The organism with this adaptation is supposed to partially destroy its kidneys on purpose in order to maintain sufficiently high blood pressure. This miocene environment is long gone. However, it is much easier to break a gene than to put it back together. Our urate oxidase gene has been broken more than once and it would take quite a long time to fix it.

It’s kind of a crazy theory. I’m not sure I believe it but it is interesting to think about.

Some Hypotheses

During this experiment, I was eating almost exclusively meat, fish (often with bones), eggs and vegetables, plus added calories from butter, coconut oil and olive oil. The diet was grain, legume and dairy free and, as mentioned, possibly ketogenic. This would be considered by many online diet and health personalities to be a good low carb paleo diet, even though of course processed fats like butter and coconut oil are not Paleolithic foods.

So I want to discuss a few possible ways to resolve the apparent impossibility of eating this way without added salt.

Hypothesis 1: Low Carb, Low Crab, or Low Salt: choose any two

I have been eating a low carb diet, and my experiment suggests that, in that context, low salt is not a good idea. It is possible that a healthy human diet can be either low in carbohydrates or low in salt, but not both.

A great deal of evidence suggests that ketosis was not the norm for our paleolithic ancestors (see e.g. Kuipers et. al. 2012 for a thorough review of paleolithic diet research). In fact it would have been quite a struggle for me to eat this sort of macronutrient ratio without modern refined fats such as butter and coconut oil. Or ready access to marine mammal blubber (but then again the Inuit are not my paleolithic ancestors).

In contrast to the online paleo diet scene, most low carb diet advocates seem to line up behind the recommendation for ample supplementary salt. My result accord with that clinical experience. Low carbohydrate diets are usually said to have a diuretic effect in this community, at least in the initial stages (e.g. M.R. Eades, Jenny Ruhl). It is possible that my problems were caused by the interaction between diet-induced ketosis and salt restriction, and I would have done just fine without salt if I had some more carbohydrates. This hypothesis would be straightforward to test.

In order to keep my sodium intake sufficiently low during the salt restriction phase, I had to remove shellfish such as oysters and mussels from my diet. Crab is also salty and makes for a handy pun. It seems likely that daily shellfish consumption would have pushed my sodium intake into the healthy range. While shellfish does not get much attention these days in the paleo club, there is ample support (again see Kuipers et. al.) that it was an important contributor to actual paleolithic nutrition.

Hypothesis 2: Humans must drink blood. Or eat salt.

File this one in the “teen paranormal romance” department. This hypothesis states that the ancestral human diet was not as low in salt as commonly assumed.

Sodium is the body’s primary extracellular cation, and most of it is located in the blood and other extracellular fluids. A pint of blood contains about 1.6 grams of sodium (see, e.g., these livestock reference ranges for blood sodium). That much blood per day should have been more than enough to push me into the healthy range of sodium consumption. On the other hand, salt depletion set in pretty quickly for me (probably 3-4 days), so this hypothesis assumes that fresh blood was consistently available to inland populations that did not have ready access to shellfish or sea water.

I find this hypothesis intriguing because of the fact that my putative ancestors were commanded not to drink blood (Genesis 9:4, Leviticus 17:13, Deuteronomy 12:15-16), and that salt is used in this tradition specifically to remove blood from meat before it is eaten. Presumably blood drinking was outlawed because it was thought to spread diseases and not because of tacky pop-culture connotations. Were my ancestors salting their meat not just for its preservative qualities, but also to make up for the reduction in sodium intake due to their prohibition on drinking blood?

Hypothesis 3: I’m Not (Genetically) a Paleolithic Human

Some say the human genome has hardly changed in the past 10,000 years. However, the hard evidence points to a number of significant evolutionary changes since the advent of agriculture, the classic example being lactase persistance (see Cochran and Harpending 2009 for a thorough argument on the rapidity of recent human evolution). Genes associated with hypertension and salt sensitivity are also apparently under strong evolutionary pressure. Alan Weder discusses this in an article published in 2007 in the journal Hypertension about evolution and hypertension. It is worth reading as an example of excellent science writing.

My experiment clearly demonstrates that I am not salt sensitive. This is not surprising given my European ancestry. As discussed by Weber, the genetics of salt resistance seem to correlate with adaptations to colder climates. It seems possible that in the course of such adaptation, my ancestors lost the ability to function optimally on a low salt diet.

Is a High Salt Diet Safe?

It is possible that, as much of mainstream medicine believes, a high salt diet actually is unhealthy over the long term. There is nothing in this experiment that contradicts that belief. Just because I am resistant to the short term blood pressure effects of salt loading, that does not mean I am immune to whatever long term negative effects a high salt diet may have. While epidemiological studies have their problems, it seems unwise to discount their findings altogether.

Edward Frohlich has argued that, notwithstanding the fact that most people’s blood pressure does not respond to acute increases in sodium intake, sodium is nevertheless responsible long-term for increases in blood pressure. He argues that excess salt causes kidney damage over time (as with uric acid this is mediated by renin/angiotensin systems), resulting long-term in an increase in blood pressure. While much of this research is based on studies done on rats (including those of the “spontaneously hypertensive” variety), this line of thought is worth looking into and I will continue to do so.

13 Replies to “Journal of Personal Science: How Much Salt Should I Eat?”

  1. Nifty stuff but the conclusions seem overstated.

    Conclusions #1 & #3 are not data-based but retrospective & subjective, and about rare things in the first place: what, do you get staph every other day? How could you draw any conclusion about rare events like that?

    #4 could have been quantified but I don’t see any graphs for it.

    #5 is reasonable but the claim about listening to your body sort of begs the question that either of the salt extremes were unhealthy enough that you should have noticed.

    #6 doesn’t mention that the weight returned to normal; when I see 2 spikes on the graph corresponding to both transitions, I wonder to myself ‘is this a Hawthorne effect where the stress of changing entirely his diet causes him to overeat a little bit?’

  2. Gwern, thanks for your comments.

    Re 1 and 3: The likelihood that these two changes (which had never happened to me before) were coincidences is much lower than the likelihood that they were caused by the intervention. Both outcomes have biological plausibility (though I have not seen prior reports associating infections with salt restriction). I would not necessarily call these “discoveries” but they are important observations that should be reported.

    Re 4: there was some level of quantification but not enough to graph in a useful manner (there are insufficient data points). This observation has more to do with how I was feeling.

    Re 5: given impaired thermoregulation, infection and strength loss, I have concluded that the low salt diet was unhealthy. However, I did not crave salt and my appetite adjusted to the lower intake. There is a body of literature on this question which I did not write up — the existence of sodium appetite in humans has been a controversial topic. It seems clear at this point that it does exist when sodium levels are critically low, but that it is extremely difficult in humans to deplete sodium to the extent necessary. My experiment did not reach the requisite level, nor should anyone attempt it without medical supervision. See the paper by Geerling and Loewy for an excellent discussion of this (“Central Regulation of Sodium Appetite”, Experimental Physiology 2008).

    Re 6: increase in extra cellular fluid is the only explanation I can think of that could describe an 8 pound change in body weight in this time frame, but I’m open to suggestions if you can think of a better one. I am fairly certain that I did not eat an extra 8 pounds of food for dinner.

  3. > Re 1 and 3: The likelihood that these two changes (which had never happened to me before) were coincidences is much lower than the likelihood that they were caused by the intervention. Both outcomes have biological plausibility (though I have not seen prior reports associating infections with salt restriction). I would not necessarily call these “discoveries” but they are important observations that should be reported.

    I don’t follow. You mean you, in advance, expected that staph and thermoregulation problems would occur and this is why you are rejecting the null? What’s the base-rate for either of these, combined with the selection effect of watching like a hawk for any change whatsoever? If you weren’t expecting in advance those specific two problems, then you are misframing the problem all the way from the start: now the right question is, what is the base rate of any out of the ordinary event at all? (Quite high – as the old point about the lottery goes, it’s surprisingly for a particular person to win, but it’s not surprising that any particular person won. See also Littlewood’s Law.)

    > Re 4: there was some level of quantification but not enough to graph in a useful manner (there are insufficient data points). This observation has more to do with how I was feeling.

    I see.

    > Re 6: increase in extra cellular fluid is the only explanation I can think of that could describe an 8 pound change in body weight in this time frame, but I’m open to suggestions if you can think of a better one. I am fairly certain that I did not eat an extra 8 pounds of food for dinner.

    I don’t see an 8lb change graphed… Where’s this 8lb stuff coming from? Your graph can’t even show an 8lb change since it only spans 6lbs, from 155 to 161.

    I see 1 datapoint for the first phase, almost useless, and variation within the second phase far exceeding variation between the second phase and third phase. I also see a second phase which is increasing just before the transition and decreasing just after the transition. Or, look at the weight graph: between the last green point and the first blue point is 2lbs. 2lbs isn’t much of a change, it looks like there are other places in the graph where 2lb shifts occur (and speaking from my personal experience, 2lb shifts happen all the time for me despite no salt experiments).

  4. I do not have a null hypothesis to reject because I was not testing a hypothesis regarding infections or thermoregulation. I’m not seeking to prove or disprove these associations. I reported these results because they are highly unusual (they never hapened before) and because they are biologically plausible. They are interesting observations, not discoveries. Why would I keep them a secret?

    The 8 pound shift is not graphed. The graph shows daily weight, measured consistently as described under Figure 3, not intra-day weight changes. Weight varies more during any given day than from day to day due to food and water consumption and excretion.

  5. Greg,
    Thx for the thoughtful and detailed note on your experiment w salt.
    I loaded salt and small amount of potassium bicarbonate and magnesium citrate into gel capsules and 2-3 of those (4-6 gms) during hiking at high altitude. I found it largely offset my loss of physical endurance at high level of exertion since going low carb.
    Other situations that this method of salt supplement helps me personally are: fasting (reducing a slight headache), preventing mental fatigue (during long distace driving).
    6 lbs of fluid retention is a lot. I never notice it personally. Is there noticeable sign?

  6. Hi there,

    When going low carb, you’d better not restrict salt or you will suffer from some quite undesirable side-effects, the most common one being constipation. I also think, IIRC, that you will start losing potassium due to salt depletion. You don’t want that …

    Anyway, as soon as you go low carb, you have to increase your intake of salt. 5g/day of sodium should do.

  7. Johnn, thanks for the comment. Salt loading at high elevation makes sense since it is easier to lose fluids up there (via respiration). Curious if you have noticed any effect of the magnesium? I took it for a while and it didn’t seem to do anything.

    The 6 pound gain was certainly noticeable, largely because I was just ending the low salt phase during which I was dehydrated. I got a moderately bad headache but felt fine in the morning.

    James, I have heard the advice to add salt during “induction” of a low carbohydrate diet but not so much emphasis on it once adapted (I’ve eaten low carb for years). It is quite possible that low salt would have worked better with more carbs in the diet, though I did not experience constipation or the other symptoms typically associated with the initial low carb adaptation period. 5g/day of sodium is quite a lot — is that recommended after adaptation to low carb and if so by whom?

  8. Hello Greg,

    Thanks for you feedback on this. A while back, I had read the book called “The Art and Science of Low Carbohydrate Living” by Phinney and Volek and I remember something about low carb – high fat diets and salt intake. I am not sure whether you have to maintain the 5g/day of sodium low carb post-adaptation but I can look it up again, I must have the book around here. FYI, I did experience a little bit of constipation at the beginning, but it is all fixed.

  9. A few thoughts
    You would like reading some of the literature where the have salt loaded the victions, I mean subjects. I think there was some food mixed in with the salt somewhere. I have read studies where they fed subjects up to 20grams of salt a day!
    Amphibians, mammals and then humans all evolved from the sea and our bodies mineral composition is similar to sea water but a bit more dilute. When we left the sea we had to carry a bit of it around inside us and balance input (diet) and output(excretion in urine and sweat).
    Salt excretion rate is influenced by insulin levels. if you insulin is low you tend to lose more salt. Conversely if you insulin is high you retain more fluid and salt. The salt hypertension link is pretty weak with experiments showing a change of only a few mmHg with large changes in salt content in diet. Some of the salt hypertension link may be a de facto meaasurement of insulin and metabolic syndrome rather than salt per se.
    Paleo style diets are very low salt- input minus output will allow balance to be maintained until one begins to exercise or is in a hot climate as salt s lost thru sweating. Peak sweat rates can result in a loss of 1-3g sodium/hour. People on a low sodium diet will sweat less salt but still significant amounts eg 1g/hour under hard working conditions. Natural meats, vegetables and fruits will amount to a similar order of magnitude with urinary loss of say 0.5 g or so per 24 hours. Hot weather and exercise tip the balance. Low carb style diet with ketosis would mean that negative sodium balance and hyponatremia is pretty much inevitable under these conditions. Persistance humters, who high heat tolerance, meant they could hunt down prey vunerable by a less developed internal heat management system. However, they would have lost a lot of salt… So, there must have been other sources of salt in the diet.
    As well as shellfish there may have been other “salt sources”” eg
    -blood consumption is known as well as liver consumption (with blood),
    -certain plants accumulate salt
    – marine species mentioned
    -mineral waters- these vary in salt content some are quite significant, traditional eskimos cooked with “öld”sea ice which was mildly salty, the waters we get from the water company generally have low mineral content.
    geophagy is in some cases from clays etc that contin significant amounts of salt but it may have other roles as well,
    many mammals\birds will use natural salt licks and early humans may have done the same- eventually social evolution may have meant they learnt to carry the salt lick around with them.
    some animals will lick each other as a cleaning and social ritual- have never heard of primitive humans doing so.
    Salt like many dietary and environmental factors seems to be something that there are certain amounts needed to achieve an optimal outcome. Many of the low carb “gurus”recommend adding 1-2 grams of salt to diet a day. Certainly worth doing during hot weather or on days when you active exercise.
    Still trying to work out a better guideline (that doesnt involve consuming blood, eating dirt or similar practices).

  10. ob, thanks for your post. I agree with everything you said. While the salt-hypertension effect is small on a population level, it can be quite large (or even slightly inverse) in individuals. There are a few studies out there showing individual responses in addition to group averages. It would benefit personal science tremendously if more studies reported data in this way.

    Do you have a good reference for the insulin/sodium connection? My insulin levels were low before going low carb, so I question that theory a bit, even though the guideline to add salt to a low carb diet is widely supported by the clinical experience of doctors working with these diets — it’s only the “why” I’m questioning.

Comments are closed.