Investigating the associations between hydration and exercise performance: methodology and limitations


RJ Maughan, Loughborough University, School of Sport, Exercise and Health Sciences, Loughborough LE11 3TU, UK. E-mail:


Loss of body water, if sufficiently severe, impairs most physiological functions, but the body water content fluctuates over the course of a normal day with no implications for physical or mental performance. The point at which an effect of dehydration becomes apparent has been the subject of much debate, in part, at least, because of the different tests that have been applied, differences in the methodologies used to induce dehydration and also because of differences in the fitness and other physiological characteristics of the subjects studied. The act of drinking itself and the conscious denial of access to water will also have implications for subjective responses to the exercise task. In many published studies, it is difficult to separate the effects of ingestion of water from those of carbohydrate, electrolytes, and other drink components. Nevertheless, there is good evidence that drinking appropriate amounts of water, especially cold water, can enhance exercise performance in many situations.


Deprivation of water for more than a few days inevitably leads to death, though survival for up to 10–14 days is possible if food intake is also absent. In the period immediately prior to death, all physiological functions will be impaired, even if normal food intake is maintained. There can be no doubt, therefore, that dehydration, if sufficiently severe, can impair both physical and mental performance. There is a strong body of experimental evidence to support the idea that a substantial reduction in the body water content (perhaps 5–10%) will reduce performance in a range of exercise tasks involving strength, power, endurance, or skilled movement.1 Equally, small fluctuations in body water content normally occur throughout the day with no perceptible effect on physical or mental performance. There is, however, considerable debate as to the effects of intermediate levels of body water loss, such as those that are likely to be incurred in daily living for some individuals, on exercise performance.2–4


The most appropriate question, perhaps, is whether the levels of water loss encountered in normal activities of daily living, including those of elite athletes, military personnel, and others exposed to hard physical exercise and climatic extremes, can cause a loss of exercise capacity or a loss of cognitive function. This may be very different from the effects observed in the artificial environment of the laboratory. In reviewing the literature, it is also essential to consider whether the study population is relevant to the target population and whether the exercise test has appropriate sensitivity and validity. In any assessment of the literature, there is also a need to consider whether any effects on performance of a reduction in body water content are a consequence of the process (dehydration) or the state (hypohydration) and to separate the effects of hypohydration from those of the methods used to induce it. These include fluid restriction, with or without food restriction; exercise, which may induce hyperthermia; and diuretic use. These methods have variable effects on electrolyte losses and this leads to differences in the distribution of water losses between the different body water compartments. There is also a need to quantify the magnitude of any hypohydration induced. This is normally based on change of body mass, with the assumption that 1 kg of mass loss is equivalent to 1 L of water loss; however, this may seriously overestimate the fluid deficit incurred in some situations.5


The validity, reliability, sensitivity, and ecological validity of tests commonly used as a measure of exercise performance have generated considerable controversy in recent years. It is clear that most laboratory and field measures used to assess exercise performance are too insensitive to detect small, but potentially meaningful, changes in performance.6 In elite sports, the margin between victory and defeat is often vanishingly small. Likewise, in the industrial environment, small errors of judgment may have serious and potentially fatal consequences in occupations such as air traffic control or in the operation of heavy machinery. Conclusions based on laboratory tests should take into account the limitations of those tests.

Until more sophisticated ergometers were developed in the 1980s, both treadmill and cycle ergometer tests usually consisted of exercise at constant speed or power that was sustained until the subject decided to terminate the test. Constant power tests to volitional exhaustion are still employed to examine the influence of various interventions on performance, but this method of testing is frequently criticized for a lack of ecological validity and poor test-retest reliability. The findings of Jeukendrup et al.,7 who found a large day-to-day variability (coefficient of variation, 27%) in time-to-exhaustion tests, and a much smaller variability in a time-trial protocol (<4%), are commonly cited to support this view. Although some continue to voice concerns over a lack of ecological validity, data from our research group report more consistent performance (coefficient of variation, 6%) in time-to-fatigue tests8 and recent reports have highlighted similar errors of measurement when changes in performance are normalized across tests.9 A key factor to consider when selecting an appropriate exercise test is its sensitivity and the smallest worthwhile effect that can be reliability detected.10 Amann et al. demonstrated that time-to-exhaustion and time-trial protocols display a similar sensitivity to the effects of hypoxia and hyperoxia on performance, and suggested that this finding will extend to other factors influencing performance.11 This is brought about by larger effects on performance in response to an intervention with constant power tests than are typically observed in time-trial protocols. This compensates for the larger test-retest variability, resulting in a very similar signal/noise ratio to that seen with time-trial protocols.10,11 In some research situations, the obvious limitation of time-trial-type testing is a difficulty in comparing the effect of an intervention on the physiological response to exercise, because at any given time one volunteer's relative power output may differ greatly from that of other participants. This can be overcome by the addition of a period of constant load exercise undertaken before the time trial, as described by Jeukendrup et al.7 However, the resulting test still does not resemble the real world of competitive sport or of the workplace. The ecological validity of time trial tests is often cited as an advantage when the intention is to apply the results to sports competitions. In reality, though, races are very different from time trials performed in isolation in the laboratory. Race pace in open competition is usually determined by the complex interaction of the leading runners rather than by an individual in isolation. Few competitors in a race have the luxury of choosing their own pace to achieve the best possible time. The one-hour record attempt in track cycling is perhaps the only competitive event that resembles the laboratory time trial, but even then, it is usual to conceal from laboratory subjects the distance covered.

A further disadvantage of time-trial tests is the need for extensive familiarization trials before experimental trials are undertaken. Even with time-to-fatigue tests, there is a need for familiarization, but adequate familiarization in these tests can usually be achieved with one or two familiarization trials in subjects who are habitually active and who are accustomed to performing strenuous exercise. Because of the need for subjects to learn effective pacing strategies, time-trial tests require a much greater number of familiarization trials. In many published studies, however, there is no mention of adequate familiarization trials having been performed, nor is there any mention of whether results were tested for the presence of an effect of trial order.8 It should also be clear that the choice of tests may be strongly influenced by the reasons for doing the studies. Studies with applications for industrial or sporting contexts should take into account the environment in which performance will occur, but attempts to better understand the underlying science may be better achieved with the more controlled conditions of the constant power-to-fatigue test.

Tests of endurance exercise performance have been more extensively studied than have tests of strength or power in high-intensity exercise. However, similar considerations apply to other exercise tests.6 A comprehensive review of the reproducibility, validity, advantages, and disadvantages of a wide range of tests of physical performance and fitness was undertaken by Saris et al.6 It is clear from the available literature that a range of different exercise tests can be used to investigate responses to hydration status. However, it is equally clear that whatever the test selected, care must be taken to standardize test conditions if valid and reproducible results are to be obtained. This includes the provision of sufficient familiarization trials to reduce any learning effects6 and an appropriate interval between tests to limit any changes in fitness or acclimation status over the time course of a study.12


Studies investigating the association between hydration status and exercise performance fall into two main categories: those in which hypohydration is induced prior to exercise and those in which hypohydration is allowed to develop as exercise progresses. The latter clearly must be of sufficient duration and intensity and the environment must be such as to induce significant levels of sweat loss. In practice, this usually means exercise lasting at least 30–60 min carried out in a warm environment. The former can include tests of strength and power as well as of endurance, and can use a range of different methods to induce hypohydration. In most cases, hypohydration is induced by a combination of exercise and heat stress, which raises the challenge of separating the effects of hypohydration and hyperthermia. It is well established that hyperthermia has a negative effect on the capacity to perform endurance exercise,13 but the effects on short-term high-intensity exercise are less clear.14 This variability might account for some of the uncertainty as to the effects of pre-exercise hypohydration on short-duration high-intensity exercise.3,4

Failure to control or account for the interactions between hydration status, body temperature, and the acute fatiguing effects of exercise may limit the interpretation of studies. Different experimental models may be equally valid, but as with exercise tests, the test parameters must be understood. Experimental models can be chosen to allow the effects of hypohydration to be separated from the acute effects of exercise and disturbances of body temperature. This can be done by inducing hypohydration the day before the experimental assessment and allowing or preventing restoration of water and electrolyte balance in the intervening period.15 A short period of rehydration to restore water deficits may or may not restore cognitive function and there are likely also differential effects on components of exercise performance.16 If the aim is to investigate the effects of an occupational or exercise stress that induces an acute loss of body water, then it may be desirable not to separate these effects. In synthesizing the available evidence, however, it is important not to mix studies carried out using different experimental models.

It is clear that there is a need for standardization of methods if sensitivity of the experimental model is to be maximized, and the following suggestions might be made. First, repeat trials should be performed on the same day of the week and should be preceded by adequate familiarization trials. Absence of a learning effect should be verified by analyzing data for a trial order effect. Second, trials should be adequately powered to give confidence in the findings. There should be standardization of eating, drinking, and exercise patterns for 48 h before each trial. Third, alcohol should be avoided and caffeine consumption standardized (and limited, if necessary, rather than avoided completely, so as to minimize any confounding effects of caffeine withdrawal) for 48 h before trials. Finally, subjects should ingest a 500-mL bolus of water 2 h prior to testing to achieve euhydration.17

When interpreting data, authors and readers should be aware of the potential for confounding effects introduced by preventing drinking when subjects wish to do so, requiring subjects to drink when they do not wish to do so, providing drinks that subjects like or do not like, providing familiar or unfamiliar drinks, and controlling effects other than those of hydration itself (carbohydrate content, temperature, etc.). It is also important to recognize that endurance performance will be affected by ambient temperature18 and humidity,19 so the choice of environment may influence whether or not an effect is seen.


In summary, the effect of hydration status on performance of various exercise tasks has been extensively studied, but uncertainties remain. This arises in large part from the limitations of the studies that have been performed and the inappropriate interpretation and application of some of the findings from these studies. In spite of the uncertainties and contradictions in the published literature, it is clear that severe reductions in body water content will impair performance and it is equally clear that athletes can benefit from intake of an appropriate amount of a well-formulated drink. The recognition, however, that each athlete is different and that general guidelines are of little value – and indeed may be dangerous – is not new,20 although repeated studies have since emphasized the interindividual variability in the sweating response to exercise, even in an apparently homogeneous population under the same exercise and environmental conditions.21


Funding.  The author's research program has been supported for many years by grants from a number of different companies with an interest in the manufacture and sale of beverages.

Declaration of interest.  The author has no relevant interests to declare.