Journal Logo


Dose Response of Caffeine on 2000-m Rowing Performance


Author Information
Medicine & Science in Sports & Exercise: March 2010 - Volume 42 - Issue 3 - p 571-576
doi: 10.1249/MSS.0b013e3181b6668b
  • Free


Evidence to date suggests that ingestion of 3-9 mg·kg−1 body mass of caffeine elicits significant improvements in endurance exercise performance (6,10-12,23,24). However, the relatively few studies that have examined the influence of caffeine on shorter-duration high-intensity exercise performance have produced inconsistent results, possibly because of significant methodological differences across studies. These differences include caffeine doses ranging from 1.3 to 13 mg·kg−1, the form of caffeine, timing of administration, subject habituation, and mode of exercise (7).

Two studies have examined the influence of caffeine on rowing performance, and both reported different findings in relation to the dose-response effect of caffeine (1,4). Bruce et al. (4) investigated the effects of ingesting a 6- and a 9-mg·kg−1 dose of caffeine or a placebo on the performance of eight well-trained male rowers in a 2000-m time trial (∼7 min in duration). They reported a 1% improvement in performance time and a 3% improvement in mean power with both 6 and 9 mg·kg−1 caffeine. In a similar study conducted on eight competitive female rowers, Anderson et al. (1) found a 1.4% increase in average power output and a 0.7% reduction in time to complete a 2000-m time trial with a moderate dose of caffeine (6 mg·kg−1) compared with placebo (P < 0.05). Supplementation with a higher dose of caffeine (9 mg·kg−1) resulted in further improvements in performance: a 2.7% increase in average power output and a 1.3% reduction in time to completion (P < 0.05). However, neither Bruce et al. (4) nor Anderson et al. (1) measured plasma caffeine concentrations, and it is not known whether there were differences in plasma caffeine levels across doses.

Caffeine consumed at doses similar to those used by Bruce et al. (4) and Anderson et al. (1) have been associated with gastrointestinal distress, dizziness, anxiety, irritability, and an inability to concentrate (19). Moreover, the minimum caffeine dose that can elicit improvements in brief high-intensity exercise is yet to be established. The aim of this study was to investigate the dose-response relationship between caffeine and 2000-m rowing performance. It was hypothesized that performance benefits would appear in a dose-response fashion such that incremental improvements would be found from 2 to 6 mg·kg−1 caffeine.



Subjects completed a V˙O2peak test, a familiarization trial, and four 2000-m time trials during a 6-wk period. Different doses of caffeine (2, 4, or 6 mg·kg−1) or a placebo were administered in a random fashion to participants 60 min before each time trial. The study was approved by an ethics committee of the University of Queensland, and all participants provided written informed consent.


Ten competitive male rowers (mean ± SD: age = 20.6 ± 1.4 yr, body mass = 87.7 ± 10.5 kg, height = 186.8 ± 6.8 cm, V˙O2peak = 5.1 ± 0.6 L·min−1) volunteered to participate in the study; all had participated in the national rowing championships in the previous year.

Preliminary testing.

Each subject reported to the laboratory on two occasions before the first time trial. On one occasion, subjects completed an incremental test to exhaustion on the rowing ergometer for the determination of V˙O2peak. On the second visit, they completed a 2000-m time trial that provided baseline performance data. Exercise in all tests was completed on a Concept II rowing ergometer (Morrisville, VT); subjects regularly trained using the Concept II rowing ergometer, and a 2000-m time trial was often used by their coach to assess their rowing ability.


Preexercise capillary blood was sampled from the earlobe (5 μL) and was immediately analyzed for lactate concentration (Lactate Pro; ARKRAY, Inc., Kyoto, Japan). V˙O2peak was established using the incremental seven-step rowing test described previously by Bourdon et al. (2). Expired air was analyzed according to the method described by Laursen et al. (18); submaximal oxygen uptakes were calculated by averaging the readings recorded during the final minute of each submaximal workload. V˙O2peak was recorded as the highest V˙O2 reading averaged during two consecutive readings, and the peak power output was recorded as the highest 30-s power output completed during the test.

The 2000-m time trial (baseline data).

One week after the initial V˙O2peak test, subjects reported to the laboratory to complete their first 2000-m time trial. The 2000-m time trial has previously been reported to be highly reliable for trained male rowers (22). The duration and intensity of the warm-up were consistent with the rower's normal competition routine. After the warm-up, subjects rested for 5 min before starting the trial.

The 2000-m time trial protocol.

Subjects completed four subsequent time trials, each separated by no less than 7 d. A 7-d interval allowed time for caffeine washout and for subjects to recover from the tests. Subjects consumed a standardized diet (200 kJ·kg−1 including 8 g·kg−1 CHO) for the 24 h before testing. As shown in Figure 1, subjects were required to report to the laboratory 90 min before testing to consume a light preexercise meal (2 g·kg−1 CHO) and for baseline HR to be measured. After consumption, a urine sample was collected, and osmolality was measured using a vapor pressure osmometer (Wescor 5500XR, Logan, UT) to assess hydration status. Venous blood (10 mL) was sampled from an antecubital vein and stored on ice for the later analysis of baseline plasma lactate, caffeine, norepinephrine, and glucose concentrations.

FIGURE 1-Caffei
FIGURE 1-Caffei:
ne supplementation test design. Meal, consumed standardized preexercise meal; VP Blood, venipuncture blood sample; HR, heart rate; Urine, urine check for hydration; Caffeine, 2, 4, or 6 mg·kg−1 anhydrous caffeine capsule; Placebo, placebo capsule; Warm-up, consistent warm-up procedure; RPE, rating of perceived exertion; 2000-m TT, 2000-m time trial; SR, stroke rate; 500-m splits, time taken to complete each 500 m; V˙O2, expired air gas analysis; Power, power output in watts; Time, total time taken to complete 2000 m; EP Blood, ear prick blood sample.

After the collection of baseline data, participants ingested either a placebo capsule (containing approximately 6 mg·kg−1 placebo-calcium sulfate) or a caffeine capsule containing 2, 4, or 6 mg·kg−1 body weight anhydrous caffeine. Capsules for all the conditions were prepared before the first trial. Caffeine was matched for color and volume with calcium sulfate so that all capsules were indistinguishable. The order of trials was randomized by a person independent of the project using a random number-generating process. Capsules were ingested 1 h before the exercise test (20). Approximately 40 min later, subjects performed a self-selected warm-up with HR and RPE collected to monitor intensity (which they were required to replicate before each of the following three time trials). Immediately after the warm-up, a further blood sample (10 mL) was taken.

Throughout each of the four 2000-m time trials, HR, stroke rate, and 500-m split times were recorded. In addition, expired air was analyzed, and power output data were recorded. The performance feedback viewed by participants was power output, average power output, and the distance remaining. Total time taken, stroke rate, HR, RPE, V˙O2, and average power output were recorded immediately after the test. Postexercise blood was sampled from an antecubital vein (10 mL) at 2 min after exercise and via ear prick (for the analysis of lactate) at 5 and 10 min after each time trial. Water was provided ad libitum during recovery. Before leaving the laboratory, subjects were asked what treatment they thought they had received (zero, small, medium, or large amount of caffeine) and whether they had experienced any side effects. Subjects were also asked at their next visit to comment on any ill effects since their previous visit.

Nutrition and training control.

Although participants were able to continue their normal training regimen throughout the study, they were asked to refrain from strenuous physical activity for 24 h immediately before each of the tests. A training diary was kept by the rowers, and this was used to monitor any significant changes to their training schedules during the experimental period.

To ensure adequate washout of caffeine, subjects were requested to abstain from caffeine-containing medications, foods, and beverages for 24 h before testing. A list of common caffeine-containing products was given to participants to assist in caffeine abstinence, and preexercise plasma caffeine was measured to assess compliance. Participants were also required to complete a questionnaire regarding their habitual caffeine consumption, which was used to assess their daily caffeine intake.

Blood sampling, storage, and analysis.

Venous blood (10 mL per sample) was sampled before supplementation, immediately before exercise, and 2 min after exercise. Samples were collected in K3EDTA (10 mL) and fluoride oxalate (2 mL) vacutainers directly from the antecubital vein, placed on ice, and, within 2 h, centrifuged at 2500-3000 rpm at 4°C for 10 min. EDTA plasma had 100 μg·mL−1 sodium metabisulfite added and was stored for norepinephrine analysis. All remaining plasma was placed in 1.5-mL storage tubes in 0.5-mL aliquots. Fluoride oxalate samples were stored in 0.3-mL aliquots, and all samples were frozen at −80°C until analysis.

Glucose, lactate, and norepinephrine analysis.

Plasma from fluoride oxalate vacutainers was measured for glucose and lactate using a method based on Trinder's (25) with glucose or lactate oxidase, respectively, on an automated laboratory analyzer (Cobas Mira, Basel, Switzerland). Blood collected from earlobe prick samples was analyzed for lactate using a portable analyzer (Lactate Pro; ARKRAY, Inc., Kyoto, Japan). The intra-assay coefficient of variation for glucose (duplicate samples) was 1.4% and that for lactate (analysis of a pooled sample containing 5.4 ± 0.1 mmol·L−1 lactate, n = 23) was 1.7%. Norepinephrine was analyzed using the method of Holmes et al. (13).

Caffeine analysis.

Plasma from K3EDTA vacutainers was used to analyze plasma caffeine concentrations. Reverse-phase high-performance liquid chromatography with ultraviolet detection at 272 nm according to the method of Koch et al. (17) with slight modifications was used for the analysis. Plasma caffeine concentrations were calculated by comparing areas under the curve, with a series of reference standards. The intra-assay coefficient of variation for caffeine (analysis of a pooled sample containing 44 ± 2 μmol·L−1 caffeine, n = 9) was 4.6%.

Statistical analyses.

A sample size calculation indicated that, to detect a 1% difference in performance (assuming 400 s to complete 2000 m = 4 s) with an SD of 4 s with α = 0.05 and power = 80%, 10 subjects would be required (paired t-test; Power and Sample Size Software, Vanderbilt University, TN).

Data were analyzed using the SPSS (version 15.0; SPSS, Inc., Chicago, IL) statistical software package. Normality of the distribution for outcome measures was tested using the Kolmogorov-Smirnov test. Analyses included standard descriptive statistics and one-way and two-way repeated-measures ANOVA. To locate the source of significant differences, the Fisher protected least significant difference (LSD) test was used. When the data were not normally distributed, the Friedman ANOVA followed by the Wilcoxon signed-rank test was used. To determine whether plasma caffeine concentration was associated with performance, a regression line within each subject was fitted resulting in 10 slope coefficients, followed by the Wilcoxon signed-rank test to determine whether the coefficients were different from zero. All tests were two-tailed, and statistical significance was set at P < 0.05. Results are given as the mean ± SD, unless stated otherwise.


Performance measures.

Performance (i.e., time and power output) was within the range of national-level rowers reported in previous studies using a 2000-m ergometer time trial (14,15,21). There were no significant differences among the four trials in performance time (P = 0.249) or average power output (P = 0.265; Table 1). Moreover, there was no effect of caffeine dose across 500-m split times (dose × time interaction, P = 0.881); however, there was an effect for time (P < 0.001) with the first 500-m split time being the fastest, followed by the second, fourth, and third 500-m splits (96.4 ± 2.5, 100.9 ± 2.7, 102.3 ± 2.8, and 102.9 ± 3.0 s, respectively). There was no significant difference in average stroke rate among trials (P = 0.212).

Rowing ergometer 2000-m performance data with placebo and three doses of caffeine (n = 10).

There was no statistically significant association (P > 0.10) between plasma caffeine concentration and rowing performance. As can be seen in Figure 2, there was considerable variation in the plasma concentrations of caffeine among subjects in response to each caffeine dose (coefficient of variation: 2 mg·kg−1 = 61.9%, 4 mg·kg−1 = 58.7%, 6 mg·kg−1 = 38.8%).

Relationship between plasma caffeine concentration before exercise and rowing performance time (P > 0.10). Each line represents an individual subject. The line for each subject is the regression of the points representing the three caffeine doses.

Average V˙O2 was not different between trials (P = 0.722) and neither was postexercise RPE (P = 0.726). Peak HR significantly increased (P = 0.020) by 1.2% ± 1.1% in the 4 mg·kg−1 trial and by 2.1% ± 2.5% in the 6 mg·kg−1 trial compared with placebo.

According to the criteria used by Van Soeren et al. (27), none of the subjects in the present study habitually consumed caffeine (i.e., all subjects consumed <400 mg·d−1 caffeine). Five subjects correctly identified each of the placebo and 6 mg·kg−1, whereas only one subject correctly identified each of the 2 and 4 mg·kg−1 trials. Of these, only three subjects correctly identified two of the four trials, whereas six correctly identified one trial only. Overall, the randomization seemed successful, and there were no significant changes in the participants' training schedules during the trial period. On only three occasions of the 30 trials involving caffeine did subjects comment that they had experienced side effects of caffeine consumption, including increased alertness, difficulty sleeping, and small hand tremors.

Plasma measures.

Initial plasma caffeine levels at baseline and in the placebo trial were all <0.3 μmol·L−1, indicating that subjects had abstained from consuming caffeinated products before testing. Across trials, there was a consistent increase (P ≤ 0.001) in plasma caffeine concentration relative to dose (Table 2).

Plasma measures on arrival at the laboratory (baseline), immediately before the 2000-m time trial (Pre), or 2 min after exercise (Post) including 5- and 10-min postexercise samples for lactate.

There were no significant differences in norepinephrine across doses (P = 0.072); however, across time points, there was a progressive increase (P < 0.001). Higher levels of norepinephrine before the commencement of exercise were significantly associated with a faster performance time (ρ = −0.337, P = 0.033).

For blood glucose concentrations, there was a significant main effect for time (baseline, before, and after exercise; P < 0.001) but not for dose (P = 0.379), and there was no significant dose × trial interaction (P = 0.357). In examining the trial effect, there was no significant difference in plasma glucose concentrations among groups at baseline (P = 0.952) or before exercise (P = 0.861). However, postexercise plasma glucose concentrations were significantly different (P = 0.010) among groups, with the 4 mg·kg−1 trial 16.6% ± 17.0% higher and the 6 mg·kg−1 trial 17.1% ± 9.5% higher compared with placebo (Table 2).

Across caffeine doses, there were no significant differences in plasma lactate concentration at baseline, at 2 min after exercise, or at 5 min after exercise. However, before exercise, there was a significant difference across doses (P = 0.006), with the 6 mg·kg−1 trial approaching significance (P = 0.051) compared with placebo (Table 2). There was also a significant difference across doses 10 min after exercise (P = 0.021), with the plasma lactate concentration significantly greater after the 6 mg·kg−1 dose compared with the other conditions.


This is the first study to consider the influence of three doses of caffeine on changes in plasma caffeine concentrations and also on subsequent rowing performance. The results indicate that the different doses of caffeine (2, 4, or 6 mg·kg−1) had no effect on the 2000-m time trial rowing performance. Moreover, there was no association between plasma caffeine concentration and performance.

After the administration of caffeine, increases were noted in peak HR and postexercise blood glucose and lactate, which are consistent with the findings of others (3,9,11,24) and demonstrate a clear physiological effect of caffeine. However, plasma caffeine concentrations 50 min after intake of 6 mg·kg−1 caffeine (24.8 ± 9.6 μmol·L−1) in the present investigation were lower than the caffeine concentrations reported by others after intake of the same dose (5,12,26). These previous researchers reported plasma caffeine values in the range of 35-50 μmol·L−1 caffeine 60 min after ingestion of 6 mg·kg−1 caffeine. Even 70 min after intake of the 6 mg·kg−1 caffeine in the present study, plasma caffeine concentrations reached only 28.0 ± 8.7 μmol·L−1. These lower-than-expected plasma caffeine concentrations may explain why the caffeine dose had no significant influence on the 2000-m rowing performance. However, differences in analytical techniques may also explain the divergent plasma caffeine values.

Differences in caffeine uptake between individuals have not been taken into account in previous investigations because subjects generally undertake a performance test 60 min after caffeine ingestion (1,4). The possibility that caffeine peaks in plasma at different times for different individuals may be one reason why the present findings are somewhat in contrast to those of Bruce et al. (4) and Anderson et al. (1), who found that the 2000-m rowing performance improved by approximately 1% after caffeine supplementation. It should be noted that caffeine administered in this study was pure anhydrous caffeine; it was therefore unlikely that compounds with the potential to interfere with some forms of caffeine (coffee, tea, chocolate, etc.) would have affected the absorption rates and influenced the findings in the present study.

Several factors are known to affect caffeine absorption and clearance, and in particular, diet and the nutritional status of individuals are important determinants (16,28). The effects of consuming a meal on the absorption and appearance of caffeine in the plasma are not clear. The preexercise meal may have the potential to slow gastric emptying and delay the absorption of caffeine. In many previous studies, the postabsorptive timing of caffeine intake has not been clearly reported by the researchers, but from the limited information provided, a preexercise meal is often consumed 4-14 h before testing (1,4,5,12,26). Participants in the present study consumed a standardized meal immediately before the ingestion of caffeine. The timing of the preexercise meal relative to caffeine intake may explain why the plasma caffeine concentrations in the current study were lower than those levels found by others (5,12,20,26). Thus, the differences in postprandial timing in this study compared with the studies by Bruce et al. (4) and Anderson et al. (1) may explain the different findings.

Despite limited information relating to differences in caffeine absorption and clearance, it is reasonable to conclude that individuals are likely to metabolize caffeine differently (8). Moreover, most investigators in the field have not reported plasma caffeine concentrations of study participants after the ingestion of caffeine, and this presents a considerable challenge when considering inconsistencies within the literature (8); the dose effects of caffeine are subtle and complex and clearly require further investigation.

The present results have an important role in informing the usage of caffeine in sports and design of future studies involving caffeine supplementation. In particular, the variation in plasma caffeine concentrations between trials suggests that several factors need to be controlled. For example, plasma caffeine concentrations after caffeine intake are likely to peak at different times for different individuals. Moreover, preexercise food consumed with the caffeine will potentially reduce peak values.

In conclusion, and in contrast to the findings of others, the present study found no improvement in the 2000-m rowing performance after the intake of caffeine up to a dose of 6 mg·kg−1. Plasma caffeine concentrations after 60 min of ingestion were lower than those values reported previously by others following the same dose, and the preexercise meal may have been responsible for this. In addition, there was considerable interindividual variation in plasma caffeine concentrations in response to the same doses administered to participants. The potential influence of preexercise feeding (and the coingestion of caffeine with food) on the subsequent appearance of caffeine in plasma warrants further investigation.

No funding was received for this work. The authors thank Mr. Gary Wilson for his assistance with the preparation and analysis of plasma samples. The results of the present study do not constitute endorsement by the American College of Sports Medicine.


1. Anderson ME, Bruce CR, Fraser SF, et al. Improved 2000-meter rowing performance in competitive oarswomen after caffeine ingestion. Int J Sport Nutr Exerc Metab. 2000;10(4):464-75.
2. Bourdon PC, David AZ, Buckley JD. A single exercise test for assessing physiological and performance parameters in elite rowers: the 2-in-1 test. J Sci Med Sport. 2009;12(1):205-11.
3. Bridge CA, Jones MA. The effect of caffeine ingestion on 8 km run performance in a field setting. J Sports Sci. 2006;24(4):433-9.
4. Bruce CR, Anderson ME, Fraser SF, et al. Enhancement of 2000-m rowing performance after caffeine ingestion. Med Sci Sports Exerc. 2000;32(11):1958-63.
5. Cole KJ, Costill DL, Starling RD, Goodpaster BH, Trappe SW, Fink WJ. Effect of caffeine ingestion on perception of effort and subsequent work production. Int J Sport Nutr. 1996;6(1):14-23.
6. Doherty M. The effects of caffeine on the maximal accumulated oxygen deficit and short-term running performance. Int J Sport Nutr. 1998;8(2):95-104.
7. Doherty M, Smith PM. Effects of caffeine ingestion on exercise testing: a meta-analysis. Int J Sport Nutr Exerc Metab. 2004;14(6):626-46.
8. Graham TE. Caffeine and exercise-metabolism, endurance and performance. Sports Med. 2001;31(11):785-807.
9. Graham TE, Helge JW, MacLean DA, Kiens B, Richter EA. Caffeine ingestion does not alter carbohydrate or fat metabolism in human skeletal muscle during exercise. J Physiol. 2000;529(3):837-47.
10. Graham TE, Hibbert E, Sathasivam P. Metabolic and exercise endurance effects of coffee and caffeine ingestion. J Appl Physiol. 1998;85(3):883-9.
11. Graham TE, Spriet LL. Performance and metabolic responses to a high caffeine dose during prolonged exercise. J Appl Physiol. 1991;71(6):2292-8.
12. Graham TE, Spriet LL. Metabolic, catecholamine, and exercise performance responses to various doses of caffeine. J Appl Physiol. 1995;78(3):867-74.
13. Holmes C, Eisenhofer G, Goldstein DS. Improved assay for plasma dihydroxyphenylacetic acid and other catechols using high-performance liquid-chromatography with electrochemical detection. J Chromatogr B Biomed Appl. 1994;653(2):131-8.
14. Jürimäe J, Jürimäe T, Mäestu J, Purge P, Sööt T. Effects of increased training volume on performance, mood state and stress hormones in trained rowers. Med Sci Sports Exerc. 2002;34(5 suppl):S72.
15. Jürimäe J, Purge P, Maestu J, Jürimäe T. Physiological responses to successive days of intense training in competitive rowers. Hungarian Rev Sports Med. 2002;43(1):11-9.
16. Kall MA, Clausen J. Dietary-effect on mixed-function P450 1A2 activity assayed by estimation of caffeine metabolism in man. Hum Exp Toxicol. 1995;14(10):801-7.
17. Koch JP, ten Tusscher GW, Koppe JG, Guchelaar HJ. Validation of a high-performance liquid chromatography assay for quantification of caffeine and paraxanthine in human serum in the context of CYP1A2 phenotyping. Biomed Chromatogr. 1999;13(4):309-14.
18. Laursen PB, Shing CM, Jenkins DG. Temporal aspects of the V˙O2 response at the power output associated with V˙O2 peak in well trained cyclists-implications for interval training prescription. Res Q Exerc Sport. 2004;75(4):421-6.
19. Magkos F, Kavouras SA. Caffeine use in sports, pharmacokinetics in man, and cellular mechanisms of action. Crit Rev Food Sci Nutr. 2005;45(7-8):535-62.
20. McLean C, Graham TE. Effects of exercise and thermal stress on caffeine pharmacokinetics in men and eumenorrheic women. J Appl Physiol. 2002;93(4):1471-8.
21. Russell AP, Le Rossignol PF, Sparrow WA. Prediction of elite schoolboy 2000-m rowing ergometer performance from metabolic, anthropometric and strength variables. J Sports Sci. 1998;16(8):749-54.
22. Schabort EJ, Hawley JA, Hopkins WG, Blum H. High reliability of performance of well-trained rowers on a rowing ergometer. J Sports Sci. 1999;17(8):627-32.
23. Spriet LL, Gibala MJ. Nutritional strategies to influence adaptations to training. J Sports Sci. 2004;22(1):127-41.
24. Spriet LL, Maclean DA, Dyck DJ, Hultman E, Cederblad G, Graham TE. Caffeine ingestion and muscle metabolism during prolonged exercise in humans. Am J Physiol. 1992;262(6):E891-8.
25. Trinder P. Determination of glucose in blood using glucose oxidase with an alternative oxygen acceptor. Ann Clin Biochem. 1969;6(24):24-7.
26. Van Soeren MH, Graham TE. Effect of caffeine on metabolism, exercise endurance, and catecholamine responses after withdrawal. J Appl Physiol. 1998;85(4):1493-501.
27. Van Soeren MH, Sathasivam P, Spriet LL, Graham TE. Caffeine metabolism and epinephrine responses during exercise in users and nonusers. J Appl Physiol. 1993;75(2):805-12.
28. WalterSack I, Klotz U. Influence of diet and nutritional status on drug metabolism. Clin Pharmacokinet. 1996;31(1):47-64.


©2010The American College of Sports Medicine