Journal Logo


Caffeine during Exercise in the Heat

Thermoregulation and Fluid-Electrolyte Balance


Author Information
Medicine & Science in Sports & Exercise: January 2009 - Volume 41 - Issue 1 - p 164-173
doi: 10.1249/MSS.0b013e318184f45e
  • Free


Caffeine ingestion is commonly used as an ergogenic aid for endurance (15). However, during prolonged exercise in the heat, the use of caffeine is discouraged based on its potential negative impact on body fluid balance and heat storage. Caffeine ingestion has been reported to reduce cutaneous blood flow during exercise in thermoneutral conditions (9). In contrast, during exercise in a hot environment, caffeine ingestion does not affect cutaneous blood flow (26). During exercise in the heat, caffeine does not affect sweat rate (17,26) or tend to increase it (13). Additionally, caffeine ingestion increases resting metabolic rate (22) and could slightly increase oxygen consumption (8) and hence heat production during exercise.

If caffeine ingestion would augment metabolic rate while reducing cutaneous blood flow, it could potentially increase heat storage during exercise. During 60 min of exercise in a 25°C, 50% relative humidity environment, Falk et al. (13) found a tendency for caffeine ingestion to increase final rectal temperature (TREC = 0.3°C; P = 0.07). However, during exercise in hotter environments (29-38°C), caffeine ingestion did not affect core temperature (17,26). Of note, Millard-Stafford et al. (17) reported that the addition of caffeine to a sports drink increased TREC above the water placebo trial but not in comparison to the ingestion of sports drinks without caffeine during cycling at standardized exercise intensity.

Another reason for refraining from caffeine consumption during exercise in a hot environment is its diuretic nature, which may exacerbate dehydration and potentially lead to heat exhaustion. At rest, ingestion of a caffeinated sports drink increases urine production (29), and this diuretic effect is also observed when a caffeinated beverage is used after exercise (14). However, during submaximal prolonged exercise, several studies coincide on that the addition of caffeine to a sports drink does not significantly increase urine production (16) even when caffeine (8.7 mg[middot]kg−1 body weight) (29) is provided at high doses. Probably, exercise somehow interferes with the stimulating effects of caffeine on renal glomerular filtration rate that is observed at rest (12).

At rest, caffeine increases urine excretion of sodium (25), chloride, and potassium (4) likely by reducing electrolytes reabsorption within nephrons. Rehydrating after exercise with a drink that contains caffeine has been reported to either increase (5) or reduce (14) urine electrolyte excretion. Most of the fluid and electrolyte losses that take place during prolonged exercise in the heat are via sweating. However, to our knowledge, the effects of caffeine ingestion on sweat electrolyte excretion have not been reported. Caffeine, although not affecting the volume of urine or sweat during exercise, could increase electrolyte losses in these fluids and exacerbate exercise electrolyte deficit.

If enough dose is ingested, caffeine increases resting (24) and exercise heart rate (HR) (3). Increases in HR during exercise in the heat may reduce diastolic filling time and stroke volume (SV), exacerbating cardiovascular drift. Caffeine also has an inotropic effect raising blood pressure at rest (24) and during exercise in a thermoneutral environment (9). If the caffeine inotropic effect persists during exercise in the heat, it may help to maintain central venous pressure and to reduce cardiovascular drift when dehydration ensues. To our knowledge, the effects of caffeine ingestion on the cardiovascular drift that develops during prolonged exercise in the heat have not been investigated.

The aim of this study was to determine the effects of caffeine on thermoregulation and fluid-electrolyte losses during prolonged exercise in the heat. For this purpose, caffeine was ingested alone or combined with water or a sports drink. We hypothesized that caffeine ingestion will not diminish the beneficial effects of rehydration on thermoregulation. However, it may alter electrolyte balance due to increased electrolyte losses in sweat and urine.



Seven endurance-trained males, who routinely trained for approximately 2 h[middot]d−1, 4-7 d[middot]wk−1, completed this study. Their mean ± SD age, weight, height, and maximal oxygen uptake (V˙O2max) were 27 ± 0.6 yr, 72 ± 7 kg, 177 ± 5 cm, and 61 ± 8 mL[middot]kg−1[middot]min−1, respectively. A physical examination (including ECG) ensured that each participant was in good health. The subjects were fully informed of any risks and discomforts associated with the experiment before giving their written consent to participate in the study. The study conforms to the Declaration of Helsinki and was approved by the local Hospital Research Ethics Committee. Subjects were light caffeine consumers (<60 mg[middot]d−1 from caffeinated soda and lyophilized coffee in milk), and all of them had previously participated in experiments involving endurance performance in hot environments.

Preliminary testing.

All participants completed a continuous incremental cycling test to volitional exhaustion to determine their V˙O2max. Each subject underwent nine consecutive days of heat acclimatization consisting of 90-min cycling bouts (63% V˙O2max) in a hot-dry environment (35°C and 27% relative humidity). The rationale behind the acclimatization was to maximize the thermoregulatory adaptations to the heat before the onset of the experimental trials. The final acclimatization bout was used to familiarize the subjects with the experimental procedures and to measure the individual's sweat rate for calculation of fluid replacement volumes during the experimental trials.

Experimental design.

A double-blind, placebo-controlled, randomized experimental design was used, with each subject serving as his own control. Each subject completed six experimental trials consisting of pedaling for 120 min (Monark® 818, Varberg, Sweden) at an intensity that initially elicited 63 ± 2% of the V˙O2max. The experimental trials were randomized and separated by at least 48 h but no more than 4 d to avoid losing heat acclimatization. The subjects received the following treatments during the exercise: 1) without rehydration (NF); 2) rehydrating 97% of their sweat losses with water (WAT); 3) with a carbohydrate-electrolytes solution (CES; Gatorade®, Quaker Oats Co., Chicago, IL); or combining these treatments with caffeine ingestion, that is, 4) CAFF + NF; 5) CAFF + WAT; and 6) CAFF + CES. All exercise trials were performed in a hot-dry environment (mean ± SD: 36 ± 1°C; 29 ± 1% relative humidity, 1.9 m[middot]s−1 airflow). Environmental temperature was recorded at 10-min intervals using a Wibget heat stress monitor (WBGT; Wibget IST, Twinsburg, OH). Maximal voluntary and electrically evoked isometric contractions of the right quadriceps were assessed before (before the caffeine ingestion) and just after the completion of exercise. After 26, 56, 86, and 116 min of exercise, maximal cycling power was measured during a 4-s all-out sprint using the inertial load method. The results of these short-term performance measurements have been recently published (10).

In trials with caffeine ingestion (i.e., CAFF + NF, CAFF + WAT, and CAFF + CES), caffeine (1,3,7-trimethylxantine, Durvitan; Seid, Barcelona, Spain) was ingested in capsules filled to provide a dose of 6 mg[middot]kg−1 of the subject's body mass. The capsules were ingested 45 min before the onset of exercise protocol because serum caffeine concentration peaks 30-60 min after ingestion (1). In trials without caffeine ingestion (i.e., NF, WAT, and CES), the subjects ingested placebo capsules filled with the same amount of dextrose. The amount of additional energy provided by the dextrose (i.e., ∼2 kcal) was deemed negligible.

Experimental protocol.

One day before each experimental trial, the subjects were instructed to adopt the same diet, light exercise, and fluid intake regimen. The subjects withdrew from all dietary sources of caffeine or alcohol for the 48 h before the test, and they reported to the laboratory at the same time of day to avoid the effects of circadian variation on the variables measured. Postprandial state was standardized by ingesting a carbohydrate meal (∼3 g carbohydrate[middot]kg−1 body mass) 3 h before the exercise. Upon arrival, subjects voided and after that they ingested 500 mL of tap water. Then, subjects rested for 2 h in a thermoneutral room (22°C) after which they voided again. A urine sample was collected in a sterilized container and its specific gravity (Usg) was immediately determined (optical refractometer; Atago, Bellevue, WA). Following subjects ingested a capsule containing caffeine (432 ± 42 mg) or placebo with 75 mL of water and rested for 45 min instrumented. After instrumentation, subjects entered the climatic chamber and sat quietly on the cycle ergometer for 5 min while resting values were obtained. Then, a 5-mL blood sample was withdrawn and subjects started pedaling.

Fluid ingestion.

In trials with fluid replacement (i.e., WAT, CES, CAFF + WAT, and CAFF + CES), subjects ingested 2.4 ± 0.1 L of fluid during exercise to replace 97% of the sweat losses. One third of the total volume was ingested right before the start of exercise and the remaining in four aliquots after 9, 30, 60, and 90 min of exercise. To prevent body cooling due to cold fluid ingestion, drinks remained for at least 45 min inside the hot chamber and their average temperature at ingestion was approximately ∽32°C.

Thermoregulatory responses.

Rectal temperature (TREC) was measured during exercise using a flexible thermistor (model 401; YSI, Dayton, OH) positioned 15 cm past the anal sphincter. Four superficial thermistors (model 409; YSI) were affixed to the skin of the leg, the thigh, the chest, and the arm to calculate mean skin temperature (23). Forearm skin blood flow (SKBF) was measured using a laser Doppler flowmeter (Moor Lab; Moor Instruments, Devon, UK) with an optical probe positioned at the dorsum of the left forearm while the arm was supported by a sling at heart level. SKBF was normalized using subject's preexercise value for each trial. All these probes were connected to a multichannel A/D board (PowerLab 8SP; ADI, London, UK) and associated software that displayed and stored data for 60 s every 10 min throughout the trials. Thermistor probes were calibrated before the study using a water bath (Vertex; Velp, Milano, Italy) and a reference high-resolution (0.1°C) mercury in-glass thermometer (Select; Proton, Madrid, Spain).

Renal responses.

Urine volume collected before, during, and immediately after exercise was measured in a graduated glass cylinder (Symax; Proton). Urine volume collected immediately after exercise was considered to be formed during exercise and thus added to the exercise volume for calculations. Urine flow (UF) before and during exercise was calculated dividing urine volume by time elapsed between collection periods (∼2 h). A urine specimen was analyzed for osmolality (OsmUrine) using a freezing point osmometer (3300; Advanced Instruments, Norwood, MA), for sodium and potassium concentrations ([Na+]Urine and [K+]Urine) using a flame photometer (PFP7-clinical; Jenway, Essex, UK), and for chloride concentration ([Cl]Urine) using an ion electrode analyzer (Easylite; Medica, Bedford, MA). Renal osmolar clearance (Cosm = UF × (OsmUrine / OsmSerum)) and free-water clearance (CH2O = UF − Cosm) were calculated to determine renal water excretion (29).

Sweat responses.

Before exercise, two sweat patches (sterilized 5 × 5-cm gauze covered with powder-free latex) were attached to the skin of the lower back using an adhesive wound dress (10 × 12 cm; Tegaderm, 3M, St. Paul, MN). The skin was cleaned with distilled-deionized water and dried with sterile gauze. Upon exercise termination, the internal gauze of the patches was collected, and sweat was separated by centrifugation (MPW-350R; MedInstruments, Warsaw, Poland). Sweat was analyzed for osmolality (OsmSweat), sodium, potassium, and chloride concentrations ([Na+]Sweat, [K+]Sweat, and [Cl]Sweat, respectively) using the same apparatus than for urine analysis. Whole-body sweat volume during exercise was estimated by subtracting preexercise from postexercise nude body weight (±0.05 kg, Wildcat; Mettler Toledo, Toledo, OH), correcting for fluid intake, urine production, and metabolic carbon and respiratory water losses (18).

Cardiovascular responses.

V˙O2 and carbon dioxide production (V˙CO2) were measured during exercise using computerized open-circuit spirometry (Quark b2; Cosmed, Rome, Italy). Cardiac output (Q˙) was determined in duplicate using a computerized version of the CO2-rebreathing technique of Collier (7) adjusting for hemoglobin concentration. HR was measured using an HR monitor (Vantage; Polar, Kempele, Finland). Stroke volume (SV) was calculated as SV = Q˙ / HR. Systolic blood pressure (SBP) and fourth phase diastolic blood pressure (DBP) were measured on the left arm using an automatic blood pressure monitor (Tango; Suntech MedInstrument, Morrisville, NC). Mean arterial pressure (MAP) was calculated as MAP = DBP + 0.33 (SBP − DBP). Systemic vascular resistance (SVR) was calculated as SVR = MAP / Q˙. Data for these variables were collected during a 6-min period beginning after 20, 55, and 115 min (Q˙ only after 20 and 115 min). V˙O2 and V˙CO2 measurements were used for calculation of net metabolic heat production (MNet = ((3.869 V˙O2) + (1.195 V˙CO2)) (4.186/60) − workload in watts) (6).

Blood sampling and analysis.

Before exercise, a 22-gauge Teflon® catheter (Insyte; Becton Dickinson, Franklin Lakes, NJ) was inserted into an antecubital vein for blood sampling. Blood samples (5 mL) were withdrawn at rest and after 8, 53, and 113 min of exercise. Blood was immediately analyzed for hemoglobin concentration (ABL-520; Radiometer, Madrid, Spain), and hematocrit was measured in triplicate by microcentrifugation (Biocen; Alresa, Madrid, Spain) after correction for trapped plasma and venous sampling. Percent changes in blood volume and plasma volume from rest were calculated with the equations outlined by Dill and Costill (11). A portion (2.5 mL) of each blood sample was allowed to clot, and serum was separated by centrifugation. The serum was analyzed for osmolality (OsmSerum; 3300, Advanced Instruments) and sodium concentration ([Na+]Serum; PFP7-clinical; Jenway). The remaining amount of blood (1.5 mL) was mixed with ethylenediaminetetraacetic acid in plastic tubes and plasma separated by centrifugation (MPW-350R; MedInstruments) and was stored at −80°C for caffeine concentration determination.

Plasma caffeine concentration.

An Agilent Technologies HPLC 1200 system (Santa Clara, CA) was used to analyze plasma caffeine concentrations. Caffeine and β-hydroxyethyltheophylline were purchased from Sigma-Aldrich S.A. (Madrid, Spain). For each 1000 mL of mobile phase acetonitrile, tetrahydrofuran, glacial acetic acid, and distilled water were mixed in the following proportion; 20:20:5:955 v/v/v/v (Merck S.A. Barcelona, Spain). The solution was degassed with a vacuum degasser and was pumped isocratically at a flow rate of 1 mL[middot]min−1. Separation was performed at room temperature on a reversed phase C18 column (4 × 150 mm, 5-μm particle size; ACE, Aberdeen, Scotland). Spectrophotometric detection was carried out at 273 nm. Chromatographs were analyzed using the ChemStation software (Agilent Technologies).

Plasma caffeine concentration from samples obtained at rest and after 8 and 113 min of exercise was analyzed. Fifty microliters of the internal standard (20 μg[middot]mL−1) were added to 50 μL of plasma and vortex mixed. Then 20 μL of 20% perchloric acid (Sigma-Aldrich) was added and the sample centrifuged at 2000g for 5 min. The supernatant was then directly applied to the HPLC column.

Statistical analysis.

Data collected before treatments (i.e., preexercise body mass, urine specific gravity, and serum osmolality) were analyzed with one-way ANOVA with repeated measures to determine whether subjects differed in their initial hydration status. Differences between trials in sweat volume were also analyzed using one-way ANOVA. Data collected repeatedly over time were analyzed using two-way (time × trial) repeated-measures ANOVA. After a significant F-test (Greenhouse-Geisser adjustment for sphericity), pairwise differences were identified using Tukey's significance (HSD) post hoc procedure. To test the main effect of caffeine, data from trials with caffeine ingestion (CAFF + NF, CAFF + WAT, and CAFF + CES) were pooled and compared with data from trials without caffeine ingestion (NF, WAT, and CES) using paired t-test. The significance level was set at P < 0.05. Figures are illustrated as means ± SEM for clarity of presentation, and all other data are presented as means ± SD.


Fluid balance.

Subjects began each trial in a similar hydration state as evidenced by a no significant difference in preexercise body mass, OsmSerum, and Usg (Table 1). During the 120 min of exercise in the heat without fluid replacement (NF trial), subjects lost 2.3 ± 0.5 L by sweating. The ingestion of caffeine (CAFF + NF trial) did not affect the amount of sweat lost (2.3 ± 0.5 L; Table 2) during exercise. Rehydration during exercise (i.e., WAT and CES trials) increased sweat losses (2.5 ± 0.4 L; P < 0.05; Table 2). The combination of caffeine ingestion with rehydration (CAFF + WAT and CAFF + CES) did not affect sweat losses (Table 2).

Preexercise body mass, serum osmolality (OsmSerum), and urine specific gravity (U sg).
Fluid balance during 120-m cycling in a hot environment at 63% V˙O2max; without rehydration (NF), rehydrating 97% of sweat losses with water (WAT), with a carbohydrate-electrolyte solution (CES), or combining these treatments with caffeine ingestion (CAFF + NF, CAFF + WAT, and CAFF + CES).

Thermoregulatory responses.

MNet increased to 2.0 ± 1.0% (414 ± 31 to 422 ± 28 W[middot]m−2; P < 0.05) during the 20- to 115-min period of exercise similarly in all trial without influence of caffeine ingestion. From a similar resting values in all trials, rectal temperature (TREC) increased to 39.4 ± 0.3°C after 120 min of exercise without fluid ingestion (NF). The ingestion of caffeine alone (CAFF + NF) did not raise TREC above the level of the NF trial, and subjects achieved a similar temperature at the end of exercise (39.4 ± 0.3°C; Fig. 1). Rehydration with WAT or CES lowered final TREC to 38.6 ± 0.3°C (P < 0.05; Fig. 1). The addition of caffeine to water (CAFF + WAT) did not affect TREC differently than WAT. However, CAFF + CES tended to have a higher TREC than CES alone (38.9 ± 0.3°C vs 38.6 ± 0.3°C; P = 0.07; Fig. 1). From resting values, SKBF increased fivefold during NF, and it remained high during the whole trial. We found no differences in SKBF among trials at any time. From 34.3 ± 0.6°C at rest, TSK increased to 35.0 ± 0.6°C (P < 0.05) after 10 min of exercise, and it remained at those values in the NF trial. This response pattern was similar in all trials.

T REC during 120-m cycling in a hot environment (36°C; 25% relative humidity) at 63% V˙O2max; without rehydration (NF),rehydrating 97% of sweat losses with water (WAT), with a carbohydrate-electrolyte solution (CES), or combining these treatments with caffeine ingestion (CAFF + NF, CAFF + WAT, and CAFF + CES). Data for seven subjects are presented as mean ± SEM. * NF and CAFF + NF different from trials with fluid ingestion (P < 0.05).

Renal responses.

Before exercise (after the ingestion of 500 mL of tap water and 2 h of resting), UF, Cosm, and CH2O were similar among all trials (2.6 ± 0.8, 3.8 ± 1.0, and −1.2 ± 0.8 mL[middot]min−1, respectively). Exercise without rehydration (NF trial and CAFF + NF) reduced UF below resting values (1.0 ± 0.8 mL[middot]min−1; P < 0.05), whereas fluid replacement with WAT, CES, and CAFF + CES maintained UF similar to preexercise values (Fig. 2). CAFF + WAT increased UF above NF trial (P < 0.05; Fig. 2). When pooling data, trials with caffeine ingestion increased UF above trials without caffeine ingestion (P < 0.05; Fig. 2, right insert). Exercise without rehydration (NF and CAFF + NF) reduced Cosm (2.0 ± 1.5 mL[middot]min−1; P < 0.05) and CH2O (−1.0 ± 0.5 mL[middot]min−1; P < 0.05) below preexercise values. In trials with rehydration, Cosm and CH2O remained at preexercise levels without differences among trials. There was no main effect of caffeine ingestion in Cosm and CH2O (Fig. 2). Compared with the NF trial, the ingestion of WAT or CAFF + WAT reduced [Na+]Urine (from 113 ± 41 to 38 ± 15 mM; P < 0.05), [K+]Urine (from 72 ± 32 to 32 ± 25 mM; P < 0.05), and [Cl]Urine (from 153 ± 47 to 83 ± 49 mM; P < 0.05). However, CAFF + NF, CES, and CAFF + CES had similar [Na+]Urine (∼85 ± 30 mM), [K+]Urine (∼60 ± 30 mM), and [Cl]Urine (∼125 ± 55 mM) than the NF trial. There was no main effect of caffeine on urine electrolytes concentration.

Renal responses for urine flow (UF), osmolar clearance (Cosm), and free-water clearance (CH2O) prior and during 120 min of exercise in the heat at 63% V˙O2max; without rehydration (NF), rehydrating 97% of sweat losses with water (WAT), with a carbohydrate-electrolyte solution (CES), or combining these treatments with caffeine ingestion (CAFF + NF, CAFF + WAT, and CAFF + CES). Data for seven subjects are presented as mean ± SEM. *Different from NF trial (P < 0.05). †Different from preexercise (P < 0.05). The right inserts display the main effects of caffeine ingestion. ‡Different from noncaffeine trials (P < 0.05).

Lower-back sweat electrolytes concentration.

The high between-subjects variability on sweat composition (>15%) probably prevented us from observing differences on regional sweat electrolytes concentration among trials (Fig. 3). However, when data were pooled to test the main effect of caffeine on regional sweat electrolytes excretion during exercise in the heat, we observed that caffeine increased [Na+]Sweat, [K+]Sweat, and [Cl]Sweat (Fig. 3, right inserts; P < 0.05). Consequently, OsmSweat was also increased by caffeine (134 ± 18 vs 151 ± 19 mOsm[middot]kg−1 H2O; P < 0.05).

Regional sweat electrolytes concentration for sodium ([Na+]Sweat,), potassium ([K+]Sweat), and chloride ([Cl]Sweat) during 120 min of exercise in the heat at 63% V˙O2max; without rehydration (NF), rehydrating 97% of sweat losses with water (WAT), with a carbohydrate-electrolyte solution (CES), or combining these treatments with caffeine ingestion (CAFF + NF, CAFF + WAT, and CAFF + CES). Data for seven subjects are presented as mean ± SEM. The right inserts display the main effects of caffeine ingestion. ‡Different from noncaffeine trials (P < 0.05).

Cardiovascular responses.

From a similar value at 20 min (∼133 ± 6), HR increased to 158 ± 6 beats[middot]min−1 at the end of the NF trial (P < 0.05). The ingestion of caffeine (CAFF + NF) did not affect HR differently than NF. In comparison to when no fluid was ingested, fluid ingestion with or without caffeine prevented the increases in HR from the 55- to 155-min period (Table 3; P < 0.05). Q˙ was similar among trials after 20 min of exercise (∼19 ± 1.1 L[middot]min−1). When rehydrating, Q˙ was maintained at the 20-min values although it decreased after 115 min of exercise in the dehydration trials with no effect of caffeine ingestion (i.e., CAFF + NF). In accordance, SV was higher after 115 min of exercise in trials with rehydration or rehydration plus caffeine in comparison to the trials without fluid ingestion (Table 3; P < 0.05). Mean arterial pressure (MAP) was reduced at the end of exercise in the NF trial (Table 3; P < 0.05), whereas the rest of the trials (including CAFF + NF) maintained MAP. Arterial pressure and Q˙ both decreased during the dehydration trials, and thus there were no differences in systemic vascular resistance (SVR) with the rehydration trials (Table 3). PV was reduced to −8.1 ± 2.8% after 20 min of exercise in all trials. When no fluid was ingested during exercise (NF and CAFF + NF trials), PV continued declining to −15.3 ± 2.8% after 115 min (P < 0.05), whereas fluid ingestion (WAT, CAFF + WAT, CES, and CAFF + CES) maintained PV at the 15-min values (−8.6 ± 2.7%). Changes in BV resembled those described for PV.

Cardiovascular responses during 120-m cycling in a hot environment at 63% V˙O2max; without rehydration (NF), rehydrating 97% of sweat losses with water (WAT), with a carbohydrate-electrolyte solution (CES), or combining these treatments with caffeine ingestion (CAFF + NF, CAFF + WAT, and CAFF + CES).

Blood samples.

Serum osmolality (OsmSerum) increased progressively from 287 ± 5 at rest to 300 ± 5 mOsm[middot]kg−1 H2O at the end of exercise in the NF trial (P < 0.05). Serum sodium ([Na+]Serum) also increased from rest to the end of exercise in the NF trial (from 138 ± 5 to 143 ± 5 mM; P < 0.05). The ingestion of caffeine (NF + CAFF trial) did not modify OsmSerum or [Na+]Serum during exercise in comparison to NF. From similar values at rest, fluid ingestion (i.e., WAT and CES) maintained OsmSerum (288 ± 5 mOsm[middot]kg−1) and [Na+]Serum (138 ± 5 mM) during exercise similar to resting values. The addition of caffeine to rehydration (CAFF + WAT and CAFF + CES) did not modify the OsmSerum and the [Na+]Serum responses. Right before exercise (45 min after ingestion of capsules), plasma caffeine concentration was approximately 0.03 ± 0.006 mM in trials with caffeine ingestion (CAFF + NF, CAFF + WAT, and CAFF + CES). Plasma caffeine concentration increased during exercise (∼0.05 ± 0.006 mM at 60 min; P < 0.05) and remained high to the end of exercise (Fig. 4). Plasma caffeine concentration was similar among those trials. The plasma caffeine concentration in trials ingesting placebo was negligible (Fig. 4).

Plasma caffeine concentration at rest (45 min after ingestion) and during 120 min of exercise in the heat at 63% V˙O2max; without rehydration (NF), rehydrating 97% of sweat losses with water (WAT), with a carbohydrate-electrolyte solution (CES), or combining these treatments with caffeine ingestion (CAFF + NF, CAFF + WAT, and CAFF + CES). Data for seven subjects are presented as mean ± SEM. †Different from preexercise (P < 0.05). ‡Trials with caffeine ingestion different from noncaffeine trials (P < 0.05).


A growing number of scientific reports support the ergogenic effects of caffeine for exercise performance (8,10,16). However, for exercise in a hot environment, caffeine use has been traditionally discouraged without ample scientific support (2). Coinciding with a previous study (26), presently the ingestion of caffeine (6 mg[middot]kg−1 body mass) before submaximal exercise in the heat (36°C, 29% humidity, 1.9 m[middot]s−1 airflow) does not have a thermogenic effect (i.e., no effect on MNet) or hindered heat dissipation (i.e., similar SKBF, TSK, and sweat rate in all trials). However, the beneficial effects of caffeine ingestion on performance should be weighted against the novel findings of this study concerning fluid-electrolyte balance and thermoregulation: 1) combining caffeine ingestion with water (CAFF + WAT) increased exercise urine production (Table 2); 2) caffeine ingestion increased sweat electrolyte excretion (Fig. 3); and 3) the ingestion of caffeine with a sports drink (CAFF + CES) tended to result in a higher TREC than CES alone.

The tendency of caffeine to reduce the effects of a sports drink on attenuating the rise in body temperature is not isolated in the literature. The combination of caffeine with sports drink has been reported to result in higher TREC than when ingesting plain water, although it was not different from the ingestion of CES alone during exercise in the heat (17). In the study of Millard-Stafford's et al. (17), the increase in TREC could be explained by the increased carbohydrate oxidation and thus heat production with CAFF + CES compared with the water trial. In our case, there were no differences in carbohydrate oxidation between the CAFF + CES and the CES trial (10), although there are data that suggest that this should had been the case (30). On the other hand, CAFF + CES did not affect our indexes of heat dissipation (SKBF, TSK, and whole-body sweat rate) differently than CES. Neither in the study of Del Coso et al. (10) nor in the study of Millard-Stafford et al. (17) did the small observed differences in TREC (0.2-0.3°C) impaired performance, and thus its biological significance is unclear.

When caffeine was ingested alone or in combination with water or sports drink, there was a tendency for increasing urine production (21%, 43%, 15%, respectively; Table 2). In consequence, when grouping the caffeine trials, a significant diuretic effect during exercise was present (28% higher than the noncaffeine trials; Fig. 2a; P < 0.05). Furthermore, the largest diuretic effect was observed when combining CAFF + WAT (Table 2). In contrast, other researchers have reported no diuretic effect of caffeine ingestion during exercise (17,29). Three differences may explain the apparent contradiction between studies. First, Wemple et al. (29) and Millard-Stafford et al. (17) replaced approximately 73% of subject's sweat losses, whereas we replaced 97%. Our higher rate of fluid ingestion may have counteracted the reduction in glomerular filtration rate that exercise induces (29). The second and important difference is that we tested the combination of water and caffeine while caffeine was added to sports drinks in the referenced studies. Seemly, the effects of water and caffeine on diuresis are synergistic. Likely, in their studies the addition of carbohydrates and electrolytes to fluid (i.e., sports drinks) may aid on fluid retention counteracting the diuretic effects of caffeine.

Third, our experimental design is unique in that it allows to group the trials with caffeine ingestion apart from the trials without caffeine ingestion, balancing out each group for the effects of hydration or carbohydrate ingestion. By using pooling data, we increased the number of observations and thus our statistical power. UF data (Fig. 2) became statistically significant only when grouping. One may argue that the absence of significance in the noncollapsed data is likely due to a type II error (acceptance of null hypothesis when it is false). Our power calculation was based in our main variable of interest (i.e., TREC), and thus we were somewhat underpowered with seven subjects for other observations with higher variability (the ones in Figs. 2 and 3). Thus, pooling data allowed us to assess the effects of caffeine ingestion on urine flux with more statistical power than previous studies.

The larger urine production during CAFF + WAT did not raise percent dehydration in comparison to the others trials with rehydration because urine represented a low percentage of total fluid losses during exercise (most of fluid lost was via sweating; Table 2). However, five subjects ought to stop cycling to void during the water ingestion trials (WAT and CAFF + WAT). The balance between the physiological benefits of full rehydration and the caffeine ingestion should be weighted against the detriment in performance caused by the time wasted to void.

On the basis of resting experiments, it could be speculated that caffeine ingestion, through its effect on elevating catecholamines (15), could elevate sweat gland activity also during exercise (27). On the other hand, the sudorific effect of adrenaline is not observed at the whole-body level (21). Although Falk et al. (13) found a trend for exercise sweat rate to be elevated by caffeine ingestion (i.e., 13%; P = 0.07), the studies conducted in the heat have not confirmed their findings (17,29). In agreement with the studies in the heat, in this experiment caffeine ingestion did not affect sweat rate when ingested alone or in combination with rehydrating drinks (Table 2). However, as a novel finding, it was observed that coffein increased lower-back sweat electrolytes concentration (10 ± 3% for sodium, 17 ± 4% for chloride, and 16 ± 5% for potassium; P < 0.05). Because sweat rate was similar in trials with and without caffeine ingestion, this effect may be related to a reduced reabsorption of electrolytes in the sweat gland duct.

Higher sweat electrolyte concentration with caffeine ingestion but similar sweat rates would result in higher total sweat sodium excretion during the caffeine trials. However, caffeine did not affect blood serum sodium concentration that was similar to the trials without caffeine. Montain et al. (20) recently calculated that an increase in sweat sodium concentration from 50 to 75 mM would not lower serum sodium in trials lasting around 2 h. The effect of caffeine ingestion on sweat electrolyte losses and its impact on serum sodium concentration during trials lasting beyond 2 h is currently unknown.

After 115 min of exercise in the NF trial, HR increased whereas Q˙, SV, and MAP declined (classical cardiovascular drift responses). Ingestion of caffeine before exercise (i.e., CAFF + NF) prevented the decline in MAP and tended to reduce HR drift. A better maintenance of blood pressure with caffeine ingestion has been previously reported when exercising in a thermoneutral environment (9). The tendency for a reduction in HR after caffeine ingestion has been shown in young children during exercise in a thermoneutral environment (28). However, most adult studies find either no change (9) or increases in exercise HR with caffeine ingestion (3). Alike previous studies (19), fluid ingestion (water or sports drinks) reduced the cardiovascular drift (better maintenance of HR, Q˙, and blood pressure) observed when no fluid was allowed. However, the combination of caffeine with fluid ingestion did not have further cardiovascular effects.

In summary, the acute ingestion of caffeine (6 mg[middot]kg−1) before 2 h of cycling in the heat did not have a measurable thermogenic effect or affected heat dissipation when ingested alone or combined with typical rehydrating beverages (water or sports drinks). However, caffeine increased urine flux during exercise mainly when combined with water (CAFF + WAT), although it did not exacerbate dehydration (low contribution of urine to total fluid losses). In addition, caffeine ingestion increased regional sweat electrolyte concentration, although serum sodium levels were not altered during 2 h of exercise.

The authors wish to thank the subjects for their invaluable contribution to the study. Partial support for this study was provided by the Gatorade Sports Science Institute. Juan Del Coso was supported by a predoctoral fellowship from the Castilla-La Mancha government in Spain. The results of the present study do not constitute endorsement by the American College of Sports Medicine.


1. Armstrong LE. Caffeine, body fluid-electrolyte balance, and exercise performance. Int J Sport Nutr Exerc Metab. 2002;12(2):189-206.
2. Armstrong LE, Casa DJ, Maresh CM, Ganio MS. Caffeine, fluid-electrolyte balance, temperature regulation, and exercise-heat tolerance. Exerc Sport Sci Rev. 2007;35(3):135-40.
3. Bell DG, Jacobs I, Zamecnik J. Effects of caffeine, ephedrine and their combination on time to exhaustion during high-intensity exercise. Eur J Appl Physiol Occup Physiol. 1998;77(5):427-33.
4. Bergman EA, Massey LK, Wise KJ, Sherrard DJ. Effects of dietary caffeine on renal handling of minerals in adult women. Life Sci. 1990;47(6):557-64.
5. Brouns F, Kovacs EM, Senden JM. The effect of different rehydration drinks on post-exercise electrolyte excretion in trained athletes. Int J Sports Med. 1998;19(1):56-60.
6. Brouwer E. On simple formulae for calculating the heat expenditure and the quantities of carbohydrate and fat oxidized in metabolism of men and animals, from gaseous exchange (oxygen and carbonic acid output) and urine-N. Acta Physiol Pharmacol Neerl. 1957;6:795-802.
7. Collier CR. Determination of mixed venous CO2 tensions by rebreathing. J Appl Physiol. 1956;9(1):25-9.
8. Cureton KJ, Warren GL, Millard-Stafford ML, Wingo JE, Trilk J, Buyckx M. Caffeinated sports drink: ergogenic effects and possible mechanisms. Int J Sport Nutr Exerc Metab. 2007;17(1):35-55.
9. Daniels JW, Mole PA, Shaffrath JD, Stebbins CL. Effects of caffeine on blood pressure, heart rate, and forearm blood flow during dynamic leg exercise. J Appl Physiol. 1998;85(1):154-9.
10. Del Coso J, Estevez E, Mora-Rodriguez R. Caffeine effects on short-term performance during prolonged exercise in the heat. Med Sci Sports Exerc. 2008;40(4):744-51.
11. Dill DB, Costill DL. Calculation of percentage changes in volumes of blood, plasma, and red cells in dehydration. J Appl Physiol. 1974;37(2):247-8.
12. Eddy NB, Downs AW. Tolerance and cross-tolerance in the human subject to the diuretic effect of caffeine, theobromine, and theophylline. J Pharmacol Exp Ther. 1928;33:167-74.
13. Falk B, Burstein R, Rosenblum J, Shapiro Y, Zylber-Katz E, Bashan N. Effects of caffeine ingestion on body fluid balance and thermoregulation during exercise. Can J Physiol Pharmacol. 1990;68(7):889-92.
14. Gonzalez-Alonso J, Heaps CL, Coyle EF. Rehydration after exercise with common beverages and water. Int J Sports Med. 1992;13(5):399-406.
15. Graham TE. Caffeine and exercise: metabolism, endurance and performance. Sports Med. 2001;31(11):785-807.
16. Kovacs EM, Stegen JHCH, Brouns F. Effect of caffeinated drinks on substrate metabolism, caffeine excretion, and performance. J Appl Physiol. 1998;85(2):709-15.
17. Millard-Stafford ML, Cureton KJ, Wingo JE, Trilk J, Warren GL, Buyckx M. Hydration during exercise in warm, humid conditions: effect of a caffeinated sports drink. Int J Sport Nutr Exerc Metab. 2007;17(2):163-77.
18. Mitchell J, Nadel E, Stolwijk J. Respiratory weight losses during exercise. J Appl Physiol. 1972;32(4):474-6.
19. Montain SJ, Coyle EF. Fluid ingestion during exercise increases skin blood flow independent of increases in blood volume. J Appl Physiol. 1992;73(3):903-10.
20. Montain SJ, Cheuvront SN, Sawka MN. Exercise associated hyponatraemia: quantitative analysis to understand the aetiology. Br J Sports Med. 2006;40(2):98-106.
21. Mora-Rodriguez R, Gonzalez Alonso J, Below PR, Coyle EF. Plasma catecholamines and hyperglycaemia influence thermoregulation in man during prolonged exercise in the heat. J Physiol. 1996;491(1):529-40.
22. Poehlman ET, Despres JP, Bessette H, Fontaine E, Tremblay A, Bouchard C. Influence of caffeine on the resting metabolic rate of exercise-trained and inactive subjects. Med Sci Sports Exerc. 1985;17(6):689-94.
23. Ramanathan N. A new weighting system for mean surface temperature of the human body. J Appl Physiol. 1964;19(3):531-3.
24. Robertson D, Frolich JC, Carr RK, et al. Effects of caffeine on plasma renin activity, catecholamines, and blood pressure. N Engl J Med. 1978;298(4):181-6.
25. Shirley DG, Walter SJ, Noormohamed FH. Natriuretic effect of caffeine: assessment of segmental sodium reabsorption in humans. Clin Sci. 2002;103(5):461-6.
26. Stebbins CL, Daniels JW, Lewis W. Effects of caffeine and high ambient temperature on haemodynamic and body temperature responses to dynamic exercise. Clin Physiol. 2001;21(5):528-33.
27. Terrada E. Effects of adrenaline on human sweating. J Physiol Soc Japan. 1966;28(1):176-83.
28. Turley KR, Bland JR, Evans WJ. Effects of different doses of caffeine on exercise responses in young children. Med Sci Sports Exerc. 2008;40(5):871-88.
29. Wemple RD, Lamb DR, McKeever KH. Caffeine vs caffeine-free sports drinks: effects on urine production at rest and during prolonged exercise. Int J Sports Med. 1997;18(1):40-6.
30. Yeo SE, Jentjens RL, Wallis GA, Jeukendrup AE. Caffeine increases exogenous carbohydrate oxidation during exercise. J Appl Physiol. 2005;99(3):844-50.


© 2009 American College of Sports Medicine