Article Text

Download PDFPDF

Protection of total body water content and absence of hyperthermia despite 2% body mass loss (‘voluntary dehydration’) in soldiers drinking ad libitum during prolonged exercise in cool environmental conditions
  1. Heinrich W Nolte1,2,
  2. Timothy D Noakes2,
  3. Bernard van Vuuren3
  1. 1ERGOnomics TECHnologies, Pretoria, South Africa
  2. 2Department of Biokinetics, Sport and Leisure Sciences, University of Pretoria, Pretoria, South Africa
  3. 3UCT/MRC Research Unit for Exercise Science and Sports Medicine, University of Cape Town, Cape Town, South Africa
  1. Correspondence to Heinrich Nolte, ERGOnomics TECHnologies, PO Box 6264, Pretoria 0001, South Africa; heinrich{at}


The extent to which humans need to replace fluid losses during exercise remains contentious despite years of focused research. The primary objective was to evaluate ad libitum drinking on hydration status to determine whether body mass loss can be used as an accurate surrogate for changes in total body water (TBW) during exercise. Data were collected during a 14.6-km route march (wet bulb globe temperature of 14.1°C ). 18 subjects with an average age of 26±2.5 (SD) years participated. Their mean ad libitum total fluid intake was 2.1±1.4 litres during the exercise. Predicted sweat rate was 1.289±0.530 l/h. There were no significant changes (p>0.05) in TBW, urine specific gravity or urine osmolality despite an average body mass loss (p<0.05) of 1.3±0.45 kg during the march. Core temperature rose as a function of marching speed and was unrelated to the % change in body mass. This suggests that changes in mass do not accurately predict changes in TBW (r=−0.16) because either the body mass loss during exercise includes losses other than water or there is an endogenous body water source that is released during exercise not requiring replacement during exercise, or both. Ad libitum water replacement between 65% and 70% of sweat losses maintained safe levels of hydration during the experiment. The finding that TBW was protected by ad libitum drinking despite ∼2% body mass loss suggests that the concept of ‘voluntary dehydration’ may require revision.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


The extent to which humans need to replace their fluid losses during exercise remains contentious despite more than 60 years of focused research.1,,10 Prior to the 1970s, athletes were encouraged to avoid drinking during exercise because it was believed that fluid ingestion impaired exercise performance.11 This is despite the evidence collected in the classic Nevada desert studies,1 which showed that the soldiers benefited from ingesting fluids during 8-h marches in the desert. At that time, drinking guidelines for the US military prescribed only the need to drink within a daily range of fluid intakes, for example, from 4 to 12 l/day in hot environments.12 13

However, after the mid-1990s, guidelines were introduced that encouraged soldiers to drink up to 1.8 litres of fluid for each hour that they were on active duty in hot environments (wet bulb globe temperature (WBGT) >30°C).13,,15 These guidelines, which mirrored those of the American College of Sports Medicine (ACSM),16 invoked the theory that only by maintaining the pre-exercise body mass could exercisers ensure that they would not suffer an impaired exercise performance or risk the development of heat illness.14 15

The adoption of these guidelines had two consequences. First, the prevalence of exercise-associated hyponatraemia (EAH) was increased,17 18 sometimes associated with an encephalopathy (EAHE) that proved fatal in a small number of US soldiers19 and marathon runners.11 Subsequently, those guidelines were revised with the result that the incidence of EAH and EAHE in the US military has fallen substantially17 as it has among marathon runners and other endurance athletes (personal communication, Noakes, 2010). Second, these guidelines required soldiers to carry more water, which led to an increased load while on active duty.20 21

Although it is now accepted that exercisers should not be encouraged to drink ‘as much as tolerable’, there is still no consensus on the optimum rate of fluid ingestion during exercise. The recently modified ACSM guidelines advise that exercisers should drink sufficiently to ensure that their body mass loss during exercise is less than 2%.9 Body mass losses of up to 2% are often regarded as ‘voluntary dehydration’. This term refers to the observation that humans do not voluntarily drink as much water as believed to have been lost (using body mass losses as a measure) even when water is readily available. Some believe that voluntary dehydration occurs because the thirst mechanism is an inadequate stimulus to drinking.1 4 7 Others6 20 22 23 argue that drinking to the dictates of thirst is the biologically appropriate behaviour that optimises performance and prevents heat illness regardless of the exact level of dehydration that develops during exercise. This debate has special relevance for the military because soldiers ingesting fluid ad libitum will always drink less than those who are forced to drink in order to lose less than 2% of their body mass during exercise.

Accordingly, the primary objective of this study was to evaluate the effect of an ad libitum fluid replacement strategy on selected hydration status markers to determine whether this approach could prevent ‘dehydration’. In addition, we wished to determine whether body mass loss can be used as an accurate surrogate measure of the changes in total body water (TBW) during exercise. Many scientists21 24,,27 conclude that some degree of body mass loss can be explained by losses other than water so that some degree of body mass loss (not replacing all body mass losses through fluid intake) is essential to maintain normal plasma osmolality especially during prolonged exercise. This is because the POsm and not the body mass is the homeostatically regulated variable during exercise.28 Others have argued since World War II that the body contains a 2-litre fluid excess that can be lost before there are any effects of ‘dehydration’.2,,4

Thus, the scientific hypothesis that we tested was that ad libitum fluid replacement is effective in protecting TBW despite body mass loss during prolonged exercise in cool environmental conditions because this method of drinking is driven to maintain plasma and tissue osmolality.

Material and methods

Ethical clearance for this study was obtained from the Research Ethics Committee from the South African Military Health Services of the South African National Defence Force (SANDF). All subjects were required voluntarily to read and sign an informed consent.

Subject selection

All soldiers taking part in an official military exercise (N=∼250) were eligible for the study; 20 soldiers were invited to volunteer for this study. Table 1 lists the subjects' characteristics (N=18). Three days prior to the route march, the submaximal oxygen consumption of each subject was directly measured using a MetaMax portable gas analyser (Cortex Biophysik, Leipzig, Germany) during the Harvard-graded step-up test. Their predicted aerobic capacity was calculated from the submaximal exercise test results and the resulting individual regression equation for heart rate versus oxygen consumption at different workloads according to the method described in ISO 8996.

Table 1

Summary of change in body mass, TBW, USG and UOsm during exercise

Exercise intervention

The exercise intervention took the form of a competitive route march of 14.6 km. Each individual had to carry a mass of 26.5 kg including 4 litres of water, a rifle and a bush hat or cap. Participants were dressed in standard issue SANDF combat dress. Water was available for additional replenishment along the route. The subjects drank according to the dictates of their thirst (ad libitum) during the march. The core body temperature of the subjects was recorded at 1-min intervals with a CorTemp 2000 (HQ, Palmetto, Florida, USA) ambulatory remote sensing system. Core temperature data were evaluated for the potential confounding effect of fluid ingestion invalidating the ingestible sensor.29 30 The ambient temperature, wind speed, relative humidity and solar radiation were recorded for the duration of the experiment (WBGT Temperature Measurement System, QUESTEMP; Quest Technologies, Randburg, South Africa).

Hydration markers

Prior to the exercise intervention, each subject emptied his/her bladder and provided a urine and saliva sample for analysis of urine specific gravity (USG), urine osmolality (UOsm) and deuterium abundance (saliva). Saliva was chosen due to the ease and non-invasive nature of its collection, as well as the fact that saliva has been proven as a valid and convenient sampling medium for determining TBW through the diluted isotope technique.31,,34 Furthermore, it has been documented that enrichments of deuterium oxide in saliva and plasma samples were identical and reached a 2-h plateau after administration of an oral dose of the tracer. Determining TBW through the diluted isotope technique, than using procedures such as a bioelectrical impedance, remains the most reliable method currently available, producing lower coefficient of variation values.32 Nude body mass was obtained on a scale accurate to 0.1 kg (Scale CPW 150; Adam Equipment, Pretoria, South Africa). Deuterium oxide (99%) was used to prepare a 4% (weight-to-weight) solution with water. This solution was then used to prepare the individual deuterium oxide doses according to individual body mass (±0.05 g/kg body mass). Appropriate weighing of the dose bottle (to the nearest 0.1 g) was performed in order to determine the exact dose consumed by each participant. After a 2-h equilibration period,34,,44 a second saliva sample was collected in order to determine the pre-exercise TBW. At the completion of the exercise, each participant was provided with a towel to dry excess perspiration prior to re-weighing. A second urine sample and third saliva sample were collected and used for the determination of postexercise USG, UOsm and deuterium abundance (saliva). The participants then received their postexercise deuterium oxide dose followed by a 2-h equilibration period. We have recently verified this methodology through concurrent measurement of TBW through oxygen-18 dilution, indicating that TBW is not overestimated by the second deuterium oxide dose (personal communication, Nolte, 2009). Urine voided during this period was recorded for correction of isotope loss. A final saliva sample was collected and body mass measurement performed in order to calculate postexercise TBW. The samples were analysed by continuous flow isotope ratio mass spectrometry using a Europa Scientific ANCA-GSL and Geo 20–20 isotope ratio mass spectrometer. TBW (kg) was calculated using the preferred method of Halliday and Miller.31

Diluted isotope methods designed to measure TBW at the time of isotope administration are subject to systematic errors from water entering the body between the time of dosing and the sample collection.32 Corrections were made for the ingestion of the isotope dose, metabolic water production and the water added to the TBW pool through exchange with atmospheric moisture according to the methods of Schoeller et al.32 As most of the correction factors depend on metabolic rate, additional corrections were made for the postexercise equilibration period during which an increased metabolic rate (excess postexercise oxygen consumption), although marginal,45,,47 would augment the overestimation through increased metabolic water production. The sweat losses of the participating soldiers were calculated according to method previously described by Rogers et al.27 Respiratory water loss was calculated by the methods of Mitchell et al.48 For all calculations and estimations involving respiratory exchange ratio (RER) and oxygen consumption (VO2), we assumed that the RER averaged 0.85 and the oxygen consumption averaged 65% of VO2max throughout the exercise.27 49 50

Statistical analyses

Student t tests were used to compare results as all the distributions of the paired differences were normal. A Pearson's product moment correlation coefficient was used to determine relationships between appropriate variables. Statistically significant differences were indicated by a p value of less than 0.05.51


Twenty participants volunteered for the study (16 males, 4 females). The mean stature of the male and female participants was 1.72±0.05 and 1.58±0.04 m, respectively. The mean predicted VO2max of the male and female participants was 45±10.3 and 37.8±2.3 ml/kg/min, respectively. Two men failed to provide sufficiently large saliva samples required for deuterium abundance analysis and their data were therefore excluded from all analyses. The mean WBGT index value for the duration of the exercise period was 14.1°C.

Body mass loss, fluid intake and sweat rates

Body mass was reduced significantly (p<0.05) during exercise; on average, the group lost 1.3±0.5 kg (table 1). The group consumed on average 850±594 ml/h during the march (table 1). The mean sweat rate was 1289±530 ml/h (table 1). There was no significant relationship (p>0.05) between exercise time and rates of fluid intake (r=−0.13), changes in % body mass (r=0.18) and sweat rates (r=−0.25). Figure 1 indicates the total fluid intake, hourly fluid intake and predicted hourly sweat rates.

Figure 1

Mean total fluid intake, hourly fluid intake and predicted hourly sweat losses.

USG and UOsm

Neither the USG nor the UOsm changed significantly during the exercise period (table 1). Figure 2A,B shows that there was no significant relationship between either USG or UOsm and TBW at the end of exercise or between changes in these variables during exercise (figure 2C,D).

Figure 2

The relationship between (A) postexercise TBW and USG (p>0.05, r=0.53); (B) postexercise TBW and UOsm (p>0.05, r=0.15); (C) TBW and USG change (%) during exercise (p>0.05, r=−0.44); and (D) TBW and UOsm change (%) during exercise (p>0.05, r=0.18).

Core temperature measurements

The average peak core temperature of the soldiers during the exercise was 38.9±0.3°C, whereas the highest individual peak core temperature was 39.6°C. Figure 3A shows that there was no relationship between the core temperature and change in body mass during exercise. However, there was a significant positive linear relationship between total fluid intake and core temperature during exercise (figure 3B). This could be explained by significant relationships between finishing time (marching speed) and core temperature (figure 3C) as well as sweat rate (figure 3D).

Figure 3

The relationship between (A) core temperature and changes in body mass (p>0.05, r=0.18); (B) core temperature and total water intake (p<0.05, r=0.56); (C) core temperature and marching speed (p<0.05, r=0.58); (D) marching speed and sweat rate (p>0.05, r=0.29); and (E) fluid intake and sweat rate (p<0.05, r=0.96).

TBW measurements

Table 1 presents the pre and postexercise TBW results. Mean TBW did not change despite 1.3±0.5 kg body mass loss. Figure 4 shows that changes in these variables were unrelated. Of the 18 subjects tested, all lost between 0.7 and 2.4 kg body mass. Yet, TBW increased in nine, stayed the same in two and was reduced in six subjects.

Figure 4

The relationship between body mass change and TBW change (p>0.05, r=−0.16).


The first important finding of this study was that there were no significant changes to TBW, USG or UOsm in any of the subjects despite a significant (p<0.05) average body mass loss of 1.3±0.5 kg (1.98%). Thus, although the subjects developed voluntary dehydration as classically described,1 they did not show a decrease in TBW and so were not ‘dehydrated’. Instead, TBW increased marginally by about 197 g during the route march. This increase in the TBW occurred even though the mean ad libitum fluid intake was 850±594 ml/h, an important finding considering that this rate of intake was less than their hourly fluid loss. These findings suggest that changes in body mass may not accurately predict changes in TBW and raise doubts about the accuracy of using body mass losses during exercise as a surrogate marker for changes in TBW at least in the range of body mass losses that we measured.52

Maughan et al24 have recently reviewed the possible reasons why water loss alone may not explain all the mass loss during exercise. First is the production of metabolic water during fuel oxidation. A second source of water gain is the intake of exogenous water in the form of either water or the water present in food eaten during exercise. A third theoretical source is the release of water with the breakdown of muscle and liver glycogen. It has been estimated that 3–4 g of water may be complexed with each gram of glycogen stored in the liver or muscles.53 As humans can store at least 450 g of glycogen,54 55 in theory at least, 1350 g of water could be stored in this way.53 This water would become available to the TBW pool even when there is a body mass loss resulting from irreversible glycogenolysis.27 It has been calculated that an athlete who loses 2 kg of mass during a marathon race would, could in fact, be dehydrated (fluid loss) by only ∼200 g when allowance is made for the body mass loss and alternatively body water gain from these three sources.24 26 56 Calculations based on the method of Rogers et al27 predict the mean rate of carbohydrate oxidation during this study to be approximately 60±15 g/h. Considering the assumptions regarding water associated with glycogen storage and the mean duration of the route march, a mean of 436±113 g of water could have become available to our subjects during this march, excluding exogenous water as well as metabolic water production.

Although the presence of this fluid store remains hotly debated,23 24 52 the findings of this study indicate that the mean TBW was unchanged in subjects who lost an average of 1.3 kg during exercise. This was also found in our earlier study.20 This finding is therefore compatible with the presence of an endogenous body water source that is released during exercise and which therefore ‘protects’ the TBW despite a body mass loss during exercise. Whereas the source of this fluid may be uncertain,5 this does not negate the importance of our finding that up to 1.3 kg of this mass lost during exercise may not be due to this loss of water from the TBW as measured with the diluted isotope (deuterium oxide) method. Indeed this finding is compatible with the theory first proposed by Ladell2,,4 during and after World War II. In particular, Ladell2,,4 observed, as did we, the loss of 2 kg body mass prior to any urine effects becoming noticeable.

There are a number of other findings in the literature that are compatible with this interpretation. Thus, Astrand and Saltin57 found that another indicator of hydration status, plasma volume, increased during an 85-km ski race despite an average 5.5% decrease in body mass. Colt et al58 found that TBW increased by 2.4% during a 16-km foot race in subjects who lost 2.3% of body mass. In addition, Speedy et al25 showed that serum [Na+] was maintained despite an average body mass loss of 2.5 kg during a 226-km Ironman triathlon. Laursen et al59 also found that changes in body mass were unrelated to core temperature, plasma [Na+] or USG at the completion of another 226-km Ironman triathlon. Thus, all these studies show that significant loss of body mass, perhaps between 1% and 3%, may be incurred without the development of significant dehydration.24

However, we and others have also shown that much greater mass losses occur in successful athletes who drink ad libitum during prolonged exercise.25 60,,63 Thus, we do not exclude the possibility that body mass losses greater than 2% may also not carry adverse physiological effect in those who drink according to the dictates of their thirst during prolonged exercise.8

Our second important finding was that there was no relationship between % body mass loss and the exercise peak core temperature as now frequently reported.61 64 65 The core body temperature of all the participants rose steadily during exercise but did not exceed 40°C let alone 42°C, which is considered the danger level for core body temperature resulting in serious health consequences.66 Instead, core temperatures were homeostatically regulated within a normal range unrelated to the degree of mass loss during exercise. Cheuvront and Haymes67 and Byrne et al64 have reported similar findings in athletes drinking ad libitum during exercise. Paradoxically, the subject who developed the highest core body temperature (39.6°C) in this study was also the subject who consumed fluid at the highest rate (1800 ml/h), thus presenting with the lowest level of dehydration. This subject lost 0.7 kg body mass while replacing 89% of his high sweat losses (2026 ml/h). He was also among the first group of finishers (figure 3C). The finding that race winners in endurance events are usually both the hottest and the most dehydrated is frequently observed,68,,70 yet infrequently acknowledged.

Third, we found no relationship between changes in TBW and urinary markers of ‘dehydration’ as also reported by others.71 72 This conflicts with some popular guidelines73 74 but reflects the historical evidence.2,,4

Finally, we found a relationship between the rates of fluid intake and sweating (figure 3E). We are unaware of other studies that have evaluated this relationship. We do not know whether these variables are causally related or explained by their co-dependence on a third variable, for example, the exercising metabolic rate.


In summary, there were a number of important findings of this study that add to the debate on the extent to which fluid and weight losses incurred during exercise need to be replaced. Currently their position has been revised so that the newest ACSM Position Stand proposes that the mass loss during exercise should not exceed 2% of the starting body mass.9 Thus, our findings could be interpreted as supportive of that proposal because we showed that subjects maintained their pre-exercise TBW despite an average body mass loss of 1.3 kg (equivalent to a 2% body mass loss) and showed no evidence for any homeostatic failure because urinary osmolality and core body temperatures did not change substantially. However, this conclusion could also be an artefact of the study design in which subjects lost only ∼2% body mass during the exercise intervention. Two questions are left unanswered by our study. Firstly whether higher levels of body mass loss during more prolonged exercise are also associated with unchanged TBW.62 Secondly, whether large changes in body mass cause changes in TBW that is associated with deleterious physiological changes such as significant reductions in plasma volume and/or electrolyte imbalances.

Regardless, our study raises questions about the validity of the term voluntary dehydration that was first coined more than 60 years ago. Additional studies during more prolonged exercise in which athletes undergo greater changes in body mass are required to determine whether voluntary dehydration does indeed occur in those who ingest fluids ad libitum during more prolonged exercises. Indeed this study invites a more thorough interrogation of the use of the term ‘dehydration’, which should be used only when there is a proven reduction in TBW and not, as this study shows, merely a reduction in body mass during exercise.


The authors would like to thank the Director Technology Development, Department of Defence, South Africa.



  • Funding TDN is funded by the University of Cape Town, Medical Research Council and Discovery Health and BvV by the University of Pretoria.

  • Competing interests None.

  • Ethics approval Ethics approval was provided by the Research Ethics Committee of the South African Military Health Services (SAMHS) of the South African National Defence Force (SANDF).

  • Provenance and peer review Not commissioned; externally peer reviewed.