Drink temperature influences fluid intake and endurance capacity in men during exercise in a hot, dry environment


  • Toby Mündel,

    1. Human Performance Laboratory, School of Sport and Exercise Sciences, The University of Birmingham, Edgbaston, Birmingham B15 2TT, UK
    Search for more papers by this author
  • Jenny King,

    1. Human Performance Laboratory, School of Sport and Exercise Sciences, The University of Birmingham, Edgbaston, Birmingham B15 2TT, UK
    Search for more papers by this author
  • Esther Collacott,

    1. Human Performance Laboratory, School of Sport and Exercise Sciences, The University of Birmingham, Edgbaston, Birmingham B15 2TT, UK
    Search for more papers by this author
  • David A. Jones

    1. Human Performance Laboratory, School of Sport and Exercise Sciences, The University of Birmingham, Edgbaston, Birmingham B15 2TT, UK
    2. Institute for Biophysical and Clinical Research into Human Movement, Manchester Metropolitan University, Hassall Road, Alsager, Cheshire ST7 2HL, UK
    Search for more papers by this author

Corresponding author T. Mündel: School of Sport and Exercise Sciences, The University of Birmingham, Edgbaston, Birmingham B15 2TT, UK. Email: t.mundel@bham.ac.uk


The effect of different drink temperatures on the perception of exertion and exercise endurance has not been extensively investigated. Consequently, the purpose of the present study was to examine the effect of drink temperature on fluid intake and endurance during cycling in the heat. Eight healthy, non-acclimated males (26 ± 7 years; maximum oxygen uptake, 54 ± 5 ml kg−1 min−1; mean ±s.d.) cycled to exhaustion at 34°C and at 65% of their peak aerobic power, consuming a drink at either 19°C (CON) or 4°C (COLD). Six of the eight subjects cycled for longer during COLD, with exhaustion occurring at 62 ± 4 min, compared to 55 ± 4 min for CON (P < 0.05; mean ±s.e.m.). Subjects consumed significantly more fluid during COLD compared to CON (1.3 ± 0.3 l h−1 compared to 1.0 ± 0.2 l h−1; P < 0.05). Heart rate tended to be lower by ∼5 beats min−1 during COLD, and rectal temperature during the second half of the exercise period was ∼0.25°C lower during the COLD trial; however, these trends were not significant (P= 0.08 and P= 0.07, respectively). No differences were observed between trials for ventilation, concentrations of prolactin, glucose and lactate or perceived exertion. It is concluded that a drink at 4°C during exercise in the heat enhances fluid consumption and improves endurance by acting as a heat sink, attenuating the rise in body temperature and therefore reducing the effects of heat stress.

It is well documented that a high (> 30°C) ambient temperature reduces the capacity to perform prolonged exercise when compared to exercise in a thermoneutral (10–20°C) environment (Galloway & Maughan, 1997; Nybo & Nielsen, 2001; Bridge et al. 2003). Since total carbohydrate utilization in the heat is often lower than in more temperate conditions, low levels of muscle glycogen are not thought to be the limiting factor (Galloway & Maughan, 1997; Pitsiladis & Maughan, 1999); rather, the onset of fatigue has been related directly to an increased heat stress, primarily limited by thermoregulatory and fluid balance factors (Maughan, 1992; Hargreaves & Febbraio, 1998).

Adequate fluid ingestion during exercise is known to prevent a fall in plasma volume, stroke volume, cardiac output and skin blood flow, lower rectal temperature and the perception of effort and prevent a progressive rise in heart rate (see Noakes, 1993). Several interventions have been shown to be successful in improving exercise capacity and reducing thermoregulatory strain in the heat, including acclimation (e.g. Nielsen et al. 1993), whole-body precooling (see Marino, 2002) and fluid ingestion (see Kay & Marino, 2000). However, adequate acclimation requires a minimum of 1 week, and precooling requires time and equipment which may be equally restrictive. Therefore, fluid replacement is the most practical intervention for individuals exercising in the heat. Several studies have investigated the effects of different regimes of fluid ingestion on the thermoregulatory, cardiovascular and metabolic responses during exercise hyperthermia (for review, see Kay & Marino, 2000) but the majority have concentrated on the composition of rehydration solutions.

It seems surprising, given the simplicity of fluid replacement as an intervention during exercise, that the temperature of drinks ingested during exercise has not been investigated more extensively. Sandick et al. (1984) studied the effects of a range of water temperatures (5, 16, 22 and 38°C) ingested after exercise under thermoneutral conditions and found that subjects consumed the largest volumes of water at 5°C and judged water at this temperature highest on a nine-point hedonic scale of preference. Wimer et al. (1997) observed a significant attenuation in the rise of rectal and skin temperatures and whole-body sweat rate during recumbent cycling at 51% maximum oxygen consumption inline image and 26°C when cold water (0.5°C) was consumed in comparison to cool (19°C) and warm water (38°C). Costill & Saltin (1974) argued that the rate of gastric emptying was increased with the ingestion of a dilute glucose solution at 5°C when compared to 35°C, although more recent evidence casts doubt on the importance of temperature on the rate of gastric emptying (Sun et al. 1988; McArthur & Feldman, 1989).

While the limited research during exercise under temperate conditions suggests that low drink temperatures can have a beneficial effect, reducing thermoregulatory and cardiovascular strain, the evidence is surprisingly less clear during exercise under conditions of heat stress; in a study by Szlyk et al. (1989), fluid intake was enhanced and dehydration reduced, as well as lower heart rate and rectal temperatures observed during 6 h of treadmill exercise (5% incline, 4.8 km h−1, 30 min h−1) at 40°C with consumption of a 15°C drink compared to one at 40°C. However, Boulze et al. (1983) found cold water (0 and 10°C) to be more pleasurable than neutral (15 and 20°C) or warm water (30– 50°C) and intake to be lower following passive heating of the subject, and Brunstrom (2002) argued that this paradox resulted from cold water being more satiating than warm water so that less was consumed.

No published study has investigated the effects of drink temperature during the exercise regimes which are commonly used to investigate performance and capacity in the heat, i.e. moderate intensity cycling (65–75% inline image) carried out at temperatures above 30°C (e.g. Febbraio et al. 1996; Galloway & Maughan, 1997; Bridge et al. 2003). Consequently, the aim of the present study was to examine the effect of drink temperature on fluid intake and endurance during cycling in the heat (34°C). Accordingly, the hypothesis of the present study was that a cold (4°C) drink would improve exercise endurance, as measured by time to volitional fatigue, in the heat when compared to a drink at a more neutral temperature (19°C).


General design

All exercise tests were carried out on an electrically braked cycle ergometer (Lode Excalibur, Groningen, The Netherlands) set in the pedal rate-independent mode. The protocol consisted of four visits. Visit 1 was an incremental exercise test to determine inline image, max, maximal aerobic power output inline image and maximal heart rate (HRmax). Visits 2–4 involved exercising at 65% inline image to volitional fatigue in a heat chamber maintained at 33.9 ± 0.2°C and a relative humidity (RH) of 27.9 ± 0.7%. Visit 2 served to familiarize the subject with the protocol and equipment, thereby minimizing any practice effect. For the remaining two visits, subjects ingested an orange-flavoured drink maintained at either 19.4 ± 0.7°C (CON) or 3.6 ± 0.2°C (COLD). The study was carried out in a randomized, cross-over design with subjects blind to the purpose of the study.


Eight healthy, non-heat-acclimated males volunteered and provided their written informed consent to participate in the study, which was performed according to the Declaration of Helsinki and approved by the Local Ethics Committee. Subject physical characteristics were (mean ±s.d.): age, 26 ± 7 years; body mass, 81 ± 13 kg; height, 182 ± 4 cm; inline image, 54 ± 5 ml kg−1 min−1, inline image, 306 ± 39 W; HRmax, 185 ± 7 beats min−1. All subjects were familiar with cycle ergometry at this intensity and had completed a General Health Questionnaire.

Experimental design

Visit 1 Subjects performed an incremental exercise test to volitional fatigue at a self-selected cadence. The seat position, handlebar height and orientation were adjusted for each subject and the same settings were used for all subsequent rides. The initial workload was 95 W, which was increased by 35 W every 3 min until fatigue. Heart rate (HR) was monitored continuously (Polar Accurex Plus, Polar Electro Oy, Finland), as were expired O2 and CO2 (Oxycon Pro, Jaeger, Germany). The test was considered maximal if at least one of the following criteria was met: (1) final HR was within 10% of predicted maximum; (2) a clear plateau in oxygen uptake was seen; or (3) respiratory exchange ratio was equal to, or above, 1.10. Maximal aerobic power output was determined using the equation of Kuipers et al. (1985).

Visits 2–4 Subjects were advised to consume a diet high in carbohydrates in the 24 h period prior to each visit, and to minimize differences in muscle glycogen concentrations between visits they were asked to record their diet in the 24 h period before the second visit and instructed to follow the same diet before each subsequent visit. Subjects arrived at the laboratory by 09.00 h after an overnight fast, having abstained from exercise, alcohol, caffeine and smoking for the previous 24 h. Upon waking that morning, subjects were asked to consume 500 ml of water in order to arrive well hydrated. A cannula (20 gauge, Venflon, Oxford, UK) was inserted into an antecubital vein and kept patent with 0.9% saline (Baxter, Norfolk, UK) during the test. Subjects were then asked to insert a rectal thermistor 10 cm beyond the anal sphincter and empty their bladder before being weighed nude. Skin thermistors were attached to the forehead, dorsum of hand, calf and lower back whilst the subject remained seated for 40 min. A resting blood sample was taken, after which a bolus of water (8 ml (kg body weight)−1) was given. Subjects were transferred to a heat chamber (33.9 ± 0.2°C, 27.9 ± 0.7% RH) and began to exercise at a constant work rate of 65% inline image, which was maintained until exhaustion. Subjects wore shorts and a T-shirt; a fan, set at ∼0.5 m s−1 and in the same position relative to the subject for all trials, was used to circulate the air in the chamber. In the event of a subject needing to urinate during the test, the subject did so whilst remaining in the heat chamber, and the volume of urine was recorded. Exercise was stopped either at subject's volition or if their rectal temperature (Trec) exceeded 39.5°C. Following exercise, the subject was weighed nude to estimate sweat loss, which was corrected for any urinary, respiratory (Snellen, 1966) and metabolic losses (Mitchell et al. 1972), fluid intake and quantity of blood drawn, and divided by time to give an average sweat rate over the period of exercise.

Drink formulation

An orange concentrate with no added sugar (Aldi, Eire) was dissolved in water and diluted as instructed, one part concentrate to four parts water. When diluted, 100 ml of fluid provided 10 kJ (2 kcal) of energy, 0.2 g carbohydrate (sugar) and only a trace of protein and fat. The subjects were allowed fluid ad libitum during the exercise, although they were required to drink at least 300 ml of fluid every 15 min to remain euhydrated. This design allowed the measurement of fluid preference as a behavioural outcome in an attempt to clarify whether a cold fluid enhances (Szlyk et al. 1989) or reduces (Boulze et al. 1983) fluid intake in a hot environment. For CON, bottles were kept outside the chamber where ambient temperature was constant at 19.4 ± 0.7°C. For COLD, bottles were immersed in ice to ensure adequate cooling (3.6 ± 0.2°C). In order to mask the experimental hypothesis and blind the subjects as much as possible with regards to the purpose of the study, they were informed that the aim of the study was to compare the efficacy of two commercially available sports drinks.

Blood collection and analyses

Venous blood samples (8 ml) were collected into prechilled EDTA-containing tubes at rest, at 10 min intervals throughout the ride and at the point of fatigue. One millilitre of blood was separated and analysed for haemoglobin and haematocrit. The remaining whole blood was centrifuged at 2300g for 10 min at 4°C, and the plasma separated and stored at –70°C until further analysis. Haemoglobin concentration was measured using a Coulter® AC·T diff™ Analyser (Beckman Coulter Inc., Miami, FL, USA), and haematocrit was measured in triplicate by centrifugation. Changes in plasma volume were calculated from haemoglobin concentrations and haematocrit values using the equations of Dill & Costill (1974). Plasma glucose and lactate concentrations were determined enzymatically (kits from Sigma Diagnostics, UK) on a semiautomatic analyser (Cobas Bio, Basel, Switzerland). Prolactin concentration was measured using a radioimmunoassay (Skybio Ltd, Bedford, UK). Average inter- and intra-assay coefficients of variations were 5.9 and 2.7%, respectively. All hormone analyses from a single subject were carried out in the same assay batch.

Gas, temperature and HR measurement

During tests in the heat chamber, standard Douglas Bags (Cranleigh, Crowborough, UK) were used to collect expired air for 2 min at the 20 and 40 min time points and were measured with a Servomex analyser and a dry gas meter (Harvard, Kent, UK) to determine expiratory minute ventilation inline image, O2 consumption rate inline image and rate of CO2 production inline image, from which the respiratory exchange ratio (RER) was calculated. Ambient temperature was measured during each ride, and RH was calculated from the wet and dry bulb thermometer differential. The rectal and skin thermistors were connected to a Squirrel data logger (Grant Instruments, UK) and values recorded every 5 min. Weighted mean skin temperature (Tmsw) was determined using the 4-site formula of Nielsen & Nielsen (1984). Heat storage was calculated from the following equation (Havenith et al. 1995):

display math

where Tcore is the core temperature, Tskin the weighted mean skin temperature and Cb the specific heat capacity of the body tissue (3.49 J g−1°C−1). Heart rate was recorded continuously throughout by telemetry (Polar Accurex Plus, Polar Electro Oy, Kempele, Finland).

Perceived exertion

All measurements were taken every 10 min and with a minimum interval of 1 min after drink consumption. Global rating of perceived exertion (RPE) was recorded using the 15-point Borg scale (Borg, 1982).

Data and statistical analyses

All statistical analyses were carried out with the use of the SPSS package, version 12.0 (SPSS Inc., Chicago, IL, USA). Data were tested for approximation to a normal distribution. Exercise data were analysed up to 40 min and at fatigue to include the maximum number of subjects and were analysed using a two-way (time × trial) repeated measures ANOVA with a Bonferroni confidence interval adjustment when comparing main effects. Values from ANOVA were assessed for sphericity and if necessary corrected using the Huynh-Feldt method. Following a significant F test, pairwise comparisons were identified using Tukey's honestly significant difference (HSD) post hoc procedure. Total heat storage and total prolactin release were measured from the area under the curve (AUC) using the trapezoid method and corrected for basal values. For differences in time to fatigue, total prolactin release, plasma volume changes, total heat storage and fluid loss/consumption, Student's paired t tests were used. Data are reported as means ±s.e.m., unless otherwise stated. Statistical significance was accepted at P < 0.05. Statistical analysis was performed on n= 8 subjects, unless otherwise stated.


Exercise intensity and cardiorespiratory changes

Subjects cycled at very similar intensities in both trials (CON, 68 ± 2% inline image; COLD, 69 ± 2% inline image; P= 0.35). Cadence dropped from 80 ± 4 r.p.m. at 10 min to 67 ± 4 r.p.m. at fatigue for CON (P= 0.02), with corresponding figures of 84 ± 5 and 70 ± 5 r.p.m. for COLD (P= 0.02). No effect of trial was observed (P= 0.25). Six (of 8) subjects cycled for longest during COLD, with time to exhaustion occurring 11 ± 5% later than during CON; mean exercise times were 62 ± 4 min for COLD and 55 ± 4 min for CON (Fig. 1.; P= 0.04).

Figure 1.

Individual and mean ±s.e.m. exercise times for CON and COLD

The heart rate response to exercise is shown in Fig. 2. Heart rate increased steadily during exercise from 134 ± 4 (CON) and 132 ± 4 beats min−1 (COLD) at 5 min to 170 ± 5 (CON) and 165 ± 4 beats min−1 (COLD) at fatigue, a significant cardiac drift of ∼30 beats min−1 in both trials (P < 0.001). Heart rate tended to be approximately 5 beats min−1 lower during exercise for COLD than CON but this effect was not significant (P= 0.08).

Figure 2.

Heart rate response during exercise for CON (□) and COLD (•)
Rightmost data points indicate values at fatigue. *P < 0.05 significantly different from 5 min value in both trials.

No differences were found between trials for ventilation, O2 uptake or RER (all P > 0.05; Table 1), which suggests that the energy consumption, gross efficiency and relative rates of fuel oxidation were similar in both trials. There was no difference between trials for sweat rate (P= 0.4), but subjects consumed more fluid during COLD than during CON (P= 0.02; Table 1).

Table 1.  Respiratory parameters, fluid loss and consumption
  1. *P < 0.05 significantly different from CON.

inline image (l min−1)65 ± 269 ± 4
inline image (l min−1) 3.1 ± 0.2 3.1 ± 0.2
RER 0.90 ± 0.01 0.89 ± 0.01
Sweat rate (l h−1) 1.4 ± 0.1 1.4 ± 0.1
Drink rate (l h−1) 1.0 ± 0.2  1.3 ± 0.3*

Body temperature responses

Rectal temperature before exercise was the same for both trials (Fig. 3). Rectal temperature rose significantly (P < 0.001) by over 1.5°C in each condition, and there was a tendency for Trec to be lower for COLD so that by fatigue, Trec was 0.25°C lower compared to CON (38.69 ± 0.19 and 38.43 ± 0.18°C, respectively); however, this was not significant (P= 0.07). Mean weighted skin temperature rose at the start of exercise, reaching a plateau by 20 min (∼32.6°C) and declining slightly thereafter until fatigue (∼32.4°C); no effects of trial or time were observed (both P > 0.05). Total heat storage (AUC) for CON was higher than for COLD (68 ± 5 versus 57 ± 5 J °C−1 kg−1 min−1), although this effect was not significant (P= 0.13).

Figure 3.

Rectal temperature response during exercise for CON (□) and COLD (•)
Rightmost data points indicate values at fatigue. *P < 0.05 significantly different from resting value in both trials.

Plasma metabolite and hormonal responses

Concentrations of plasma lactate and glucose before exercise were the same for both trials. Plasma lactate increased from 1.2 ± 0.1 mmol l−1 at rest to 4.1 ± 0.4 mmol l−1 at 10 min in both trials (P < 0.001) and then remained relatively constant, with no differences observed between trials (P= 0.28). Plasma glucose during CON decreased from rest (5.1 ± 0.2 mmol l−1) to 4.8 ± 0.3 mmol l−1 at 10 min, with values at fatigue of 4.8 ± 0.3 mmol l−1. The corresponding values for COLD were 5.0 ± 0.2, 4.8 ± 0.2 and 5.0 ± 0.2 mmol l−1. No effects of time or trial were observed (both P > 0.05).

Circulating prolactin increased from 189 ± 47 (CON) and 167 ± 47 mU l−1 (COLD) at rest, to 322 ± 113 (CON) and 227 ± 60 mU l−1 at fatigue; however, no effects of time or trial were observed (both P > 0.05). Total prolactin release (AUC) was 6787 ± 1507 and 5562 ± 1288 mU min−1 l−1 for CON and COLD, respectively (P= 0.10). Plasma volume changes were similar in both trials (CON, –15 ± 1%; COLD, –13 ± 1%; P= 0.10), indicating that differences between the trials in plasma metabolites or hormones were probably not a consequence of haemoconcentration.

Perceived exertion

Global RPE increased significantly during exercise in both conditions (Fig. 4; P < 0.001) from values at 10 min of 11.8 ± 0.3 (CON) and 11.6 ± 0.4 (COLD) to 17.4 ± 0.4 (CON) and 17.2 ± 0.5 (COLD) at fatigue, with no effects of trial observed (P= 0.43).

Figure 4.

Perceived exertion (Borg Scale) during exercise for CON (□) and COLD (•)
Rightmost data points indicate values at fatigue. *P < 0.05 significantly different from 10 min value in both trials.


The aim of the present study was to investigate whether a cold drink (4°C, COLD) would improve exercise endurance in the heat when compared to a drink at a more neutral temperature (19°C, CON). The results of the present study suggest that fluid at 4°C is more palatable than fluid at 19°C, as demonstrated by a significantly greater intake, and that it significantly improves endurance, probably as a result of reducing the exercise-induced rise in core temperature.

In the present study, six (of 8) subjects cycled for longer during COLD, with time to exhaustion occurring 11 ± 5% later than during CON. We believe this to be the first published study to demonstrate an improved endurance capacity during exercise in the heat as a consequence of ingestion of a cold fluid. Since fluid ingestion during exercise attenuates hyperthermia, it follows that ingestion of a cold drink during exercise may induce a heat debt and thereby a lesser increase in core temperature. It has been demonstrated that ingestion of a cold fluid at rest elicits a sustained decrease in core temperature proportional to the volume and temperature of the drink (Pinson & Adolph, 1942). Gisolfi & Copping (1974) found that drinking water at 10°C reduced rectal temperature during exercise in a warm environment (33.5°C) compared to water at 37°C, whilst Szlyk et al. (1989) observed a reduced increase in rectal temperature (∼0.4°C) when consuming water at 15 compared to 40°C during and after 6 h of treadmill exercise in a hot environment (40°C). The present data are consistent with these findings, since during the second half of the exercise period and at the point of fatigue, rectal temperature was ∼0.25°C lower during COLD despite exercise time being approximately 7 min longer (Fig. 2). Although not universally accepted (Lee et al. 2004), the general consensus is that the cold fluid acts as a heat sink that cools the body (Pinson & Adolph, 1942).

Prolonged exercise in the heat is thought to be primarily limited by thermoregulatory and fluid balance factors (Maughan, 1992; Hargreaves & Febbraio, 1998). There appears to be no reduction in cardiac output, mean arterial pressure and muscle or skin blood flow (Nielsen et al. 1993) or impaired muscle function when hydration is adequately maintained, and the ability of muscle to extract oxygen seems to be well preserved (Gonzalez-Alonso et al. 1998). It has been suggested that the central nervous system may become important in the development of fatigue, particularly so when core body temperature is considerably elevated (Bruck & Olschewski, 1987; Gonzalez-Alonso et al. 1999). The implication of this work is that a high body temperature results in an inhibition of motor activity, presumably with the effect of reducing heat production and thus preventing temperature reaching dangerously high levels. Evidence for this comes from studies that have shown reduced muscle activation (Nybo & Nielsen, 2001; Martin et al. 2005) and altered brain activity as evidenced by electroencephalogram (Nybo & Nielsen, 2001; Nielsen et al. 2001) in response to exercise-induced hyperthermia. In the present study, rectal temperature at the point of fatigue in the CON trial was 38.7 ± 0.2°C and in the COLD trial it was 38.4 ± 0.2°C, values that were not statistically different (P= 0.11). This is consistent with the notion that a high core temperature is an important factor limiting performance. However, exercise continued for an average 7 min longer in the COLD condition and there was a strong trend for rectal temperature to be, on average, 0.25°C lower during the latter half of the exercise period compared to the CON trial. From the present results, it would seem that ingestion of a cold fluid attenuated the rise in core temperature, therefore increasing the time taken to reach a ‘critical’ temperature-related fatigue, as described above. Support for this comes from studies that have manipulated core temperature by precooling (see Marino, 2002), the idea being that successfully reducing body temperature increases the margin for metabolic heat production and prolongs the time to reach the critical limiting temperature at which exercise cannot be maintained (Nielsen et al. 1993). Moreover, in the present study there was a trend for heat storage to be lower for COLD, and it has previously been demonstrated that time to exhaustion in hot environments is directly related to the rate of heat storage (Gonzalez-Alonso et al. 1999).

Although the core temperatures at fatigue in the present study are not as high as the ∼40°C reported by some (Nybo & Nielsen, 2001; Nielsen et al. 2001), it has often been reported that exhaustion occurs well below 40°C (e.g. Cheung & McLellan, 1998; Bridge et al. 2003), and it seems that the interindividual variability may result from training status (Cheung & McLellan, 1998) and/or body composition (Selkirk & McLellan, 2001), although the situation is complicated by the different anatomical sites that have been used for the measurement of core temperature by the various research groups.

Since the rise in rectal temperature was attenuated with COLD, it may be expected that this would adversely affect heat dissipation by reducing global sweat rate or by decreasing evaporative heat loss. This has previously been reported by Wimer et al. (1997), who noted that ingestion of cold water (0.5°C) significantly reduced whole-body sweat loss and evaporative heat loss compared with warm water (38°C) during 2 h of recumbent cycling at 26°C. The present results do not support this suggestion, since neither whole-body sweat rate nor evaporative heat loss, as demonstrated by mean skin temperature, showed any difference between CON and COLD. In a study by Montain & Coyle (1992), subjects cycled for 2 h at ∼62% inline image and 33°C and were provided with either no fluid (NF) or 2405 ± 103 ml of a carbohydrate–electrolyte replacement solution (FR). The authors observed that when compared to NF, FR significantly attenuated the reduction in cardiac output, stroke volume and skin blood flow, as well as the rise in core temperature, although it is important to note that these differences were apparent during the second hour of exercise only. It is possible in the present study that the extra volume of fluid consumed during COLD (1330 ± 242 versus 1084 ± 236 ml; P= 0.034) could have increased skin blood flow and thus heat loss, and therefore contributed towards the improved endurance time, although it is questionable whether an additional volume of only 10% of the volume used by Montain & Coyle (1992) would have a physiological effect. Since measurements of cardiac output, stroke volume and skin (forearm) blood flow were not made in the present study, it is difficult to confirm whether a higher fluid intake per se was the reason for greater endurance.

Another consequence of COLD ingestion, compared to CON, was the attenuation of the rise in heart rate by approximately 5 beats min−1. There may have been several reasons for this: (1) cold water can cause a vagal response and produce a relative bradycardia; (2) the attenuation of the core temperature rise, in response to COLD serving as a heat sink, would decrease cardiac heat stress; or (3) greater fluid consumption may prevent dehydration to a greater extent, and therefore reduce cardiac stress. Szlyk et al. (1989) observed a HR attenuation of ∼14 beats min−1 as a result of drinking cool water at the end of 6 h of treadmill exercise under conditions of heat stress (40°C), and the authors argue that this attenuation results from a reduction in the loss of body weight and improved maintenance of plasma volume levels. In the present study, decreases in plasma volume were modest and similar in both trials, suggesting that dehydration was not a limiting factor, even though subjects consumed more fluid during COLD. It is likely, therefore, that the lower heart rate during COLD was due to either or both the first two factors discussed above.

There was no evidence in the present study that drink temperature altered substrate metabolism in any way. Concentrations of glucose and lactate were similar in both trials, and inline image and RER were similar in both trials, indicating that fuel selection and oxidation rates were not altered by drink temperature. Previous studies investigating drink temperature during and after exercise have failed to report any data pertaining to metabolism in any way (Sandick et al. 1984; Szlyk et al. 1989; Wimer et al. 1997).

When compared to CON, COLD significantly enhanced fluid consumption (Table 1). This is in agreement with the results provided by Sandick et al. (1984) but differs from the findings of Boulze et al. (1983), who reported a reduced intake of cold water (< 10°C) compared to water at 15–20°C following 15 min of passive heating at 40°C, to increase Trec by ∼0.5°C. This discrepancy probably results from the difference in protocols used and their duration. It is expected that the reason for a greater fluid intake during COLD in the present study is similar to that in the studies by Sandick et al. (1984) and Szlyk et al. (1989), who observed greater ratings of hedonic preference and reported decreased ratings of thirst when subjects consumed water at a cooler temperature. However, no such tests were administered in the present study.

Another consequence of exercise in a hot environment is a higher rating of perceived exertion (RPE) when compared to similar exercise performed in cooler environments (Pitsiladis & Maughan, 1999; Bridge et al. 2003), indicating a greater reluctance to continue exercise. In the present study, RPE increased significantly in both trials, from subjects rating their exertion as ‘light’ to values representing exertion as ‘very hard’, although drink temperature did not have any effect (Fig. 4). This finding was unexpected, since we anticipated the attenuated rectal temperature and therefore reduced heart rate during COLD to have been reflected by a lower RPE, as has been reported previously (Pitsiladis & Maughan, 1999; Bridge et al. 2003). It appears that the attenuation, although not significant, of rectal temperature and heart rate in the present study was not large enough to affect RPE. Other studies investigating drink temperature during exercise have not measured RPE (Szlyk et al. 1989; Wimer et al. 1997).

The anterior pituitary hormone, prolactin, is released into the circulation as subjects fatigue towards the end of endurance exercise (e.g. Bridge et al. 2003), and concentrations of prolactin are augmented when heat stress is combined with exercise (Brisson et al. 1991; Bridge et al. 2003). Prolactin release is controlled by secretion of inhibiting and releasing factors from the hypothalamus into the portal circulation, and a recent study using functional brain imaging (fMRI) identified a strong and significant correlation (r= 0.86) between activity in the hypothalamus and prolactin release (Anderson et al. 2002). The present results suggest that an attenuation of 0.25°C in rectal temperature during COLD was an insufficient stimulus to reduce prolactin release and/or hypothalamic activity, since prolactin concentrations during exercise and total prolactin release (AUC) were not significantly different from CON.

Limitations and considerations

At an ambient temperature of between 18 and 25°C, fluid at 19°C (CON) would be viewed as thermoneutral; however, when ambient (and skin) temperature approaches 35°C, it is possible that the relative difference in temperature between the fluid and body/environment detected was large enough to affect thermal perception so that fluid at 19°C during exercise in the present study was actually not a true control. Future studies might use a range of fluid temperatures, i.e. 5, 15 and 25°C, to explore this matter further. More cardiovascular measures, such as stroke volume and skin blood flow, would have provided insight into the possible changes in central and peripheral blood volumes and greater understanding of the specific reasons for improved exercise endurance with COLD in the present study. Finally, although the present study was sufficiently powered to investigate the primary outcome of exercise endurance, aposteriori analysis revealed that future studies should include a sample size of n= 10 to elucidate the underlying mechanisms, such as an attenuated rise in core temperature.


In conclusion, a cold fluid (4°C) significantly enhanced fluid consumption and improved exercise capacity in the heat when compared to a drink at a more neutral temperature (19°C). With cold fluid, rectal temperature was ∼0.25°C lower during the second half of the exercise period and heart rate ∼5 beats min−1 lower, suggesting that the greater volume of cold fluid acted as a heat sink, thereby reducing the effects of heat stress placed upon the body and increasing the time taken to reach an exercise-limiting core temperature.