Secondary Logo

Journal Logo

Effect of Evening Postexercise Cold Water Immersion on Subsequent Sleep


Medicine & Science in Sports & Exercise: July 2013 - Volume 45 - Issue 7 - p 1394–1402
doi: 10.1249/MSS.0b013e318287f321

Purpose This study investigated the effect of cold water immersion after evening exercise on subsequent sleep quality and quantity in trained cyclists.

Methods In the evenings (∼1900 h) on three separate occasions, male cyclists (n = 11) underwent either no exercise (control, CON), exercise only (EX), or exercise followed by cold water immersion (CWI). EX comprised cycling for 15 min at 75% peak power, then a 15-min maximal time trial. After each condition, a full laboratory-based sleep study (polysomnography) was performed. Core and skin temperature, heart rate, salivary melatonin, ratings of perceived fatigue, and recovery were measured in each trial.

Results No differences were observed between conditions for any whole night sleep measures, including total sleep time, sleep efficiency, sleep onset latency, rapid eye movement onset latency, wake after sleep onset, or proportion of the night spent in different sleep stages. Core temperature in EX and CWI trials was higher than CON, until it decreased below that of EX and CON until bedtime in CWI. After bedtime, core temperature was similar for all conditions throughout the night, except for a 90-min period where it was lower for CWI than EX and CON (3.5–4.5 h postexercise). Heart rates for EX and CWI were both significantly higher than CON postexercise until bedtime, whereas skin temperature after CWI was significantly lower than EX and CON, remaining lower than EX until 3 h postexercise. Melatonin levels and recovery ratings were similar between conditions. Fatigue ratings were significantly elevated after exercise in both CWI and EX conditions, with EX still being elevated compared with CON at bedtime.

Conclusion Whole night sleep architecture is not affected by evening exercise alone or when followed by CWI.

1School of Sport Science, Exercise and Health, The University of Western Australia, Crawley, AUSTRALIA; 2Performance Recovery, The Australian Institute of Sport, Belconnen, AUSTRALIA; 3Research Institute for Sport and Exercise Sciences, Liverpool John Moores University, Liverpool, England, UNITED KINGDOM; 4Centre for Sleep Science, School of Anatomy, Physiology and Human Biology, The University of Western Australia, Crawley, AUSTRALIA; and 5Sports Medicine, Western Australian Institute of Sport, Mt. Claremont, Western Australia, AUSTRALIA

Address for correspondence: Elisa Robey, BS (Hon.), School of Sport Science, Exercise and Health, The University of Western Australia, 35 Stirling Highway, Crawley, WA 6009, Australia; E-mail:

Submitted for publication December 2012.

Accepted for publication January 2013.

The restorative qualities of sleep for maintaining optimal bodily function are well recognized. Cognitive processes and metabolic functions, both of which are important contributors to exercise performance, can be affected by the quality and quantity of sleep (25). Indeed, sleep is considered to be a critical element of postexercise recovery procedures amongst athletes (25).

Exercise is known to influence the circadian rhythms of core temperature and melatonin secretion, which are both linked with the sleep–wake cycle (33). Exercise performed in the early evening may affect both of these rhythms; it advances the normal increase in melatonin (5) and increases core temperature (which normally peaks in the early evening) by a further approximately 1.5°C–2°C (23,28). Such an evening increase in core body temperature could potentially negatively affect subsequent sleep quality by opposing the normal decline in body temperature (postexercise, by being higher to start with) that usually precedes sleep onset. Alternatively, after exercise, it could accelerate the normal rate of decline in body temperature that precedes sleep and improve sleep propensity (19). In addition, peripheral (skin) temperature may also influence sleep quality: commonly, distal skin temperature increases in the evening as proximal skin and core temperature decline, as part of the circadian regulation of the core/shell ratio (12,13,32). The prevailing view is that high-intensity evening exercise is detrimental to subsequent sleep, and typical recommendations for optimal sleep hygiene in normal, healthy individuals commonly state that physical activity should be avoided less than 3 h before bedtime (14). However, it is unclear whether such recommendations apply to competitive athletes who are often required to undertake high-intensity evening exercise, either as part of their training regimen or because of evening competitions, as often sleep is attempted shortly after the completion of exercise or competition.

To date, few studies have investigated the effect of evening exercise on subsequent sleep in athletic populations. Myllymaki et al. (16) recently reported that cycling exercise to volitional exhaustion in the evening (∼2100 h) in young fit individuals (but nonathletes) increased non–rapid eye movement (NREM) and decreased rapid eye movement (REM) sleep, whereas Youngstedt et al. (34) found that prolonged and vigorous late night exercise in cyclists, finishing 30 min before bedtime, did not affect their subsequent sleep quality. More recently, Flausino et al. (7) reported that evening aerobic exercise improved sleep in healthy active individuals.

Athletes commonly use cold water immersion (CWI), usually in approximately 14°C water for approximately 15 min, as a postexercise recovery technique (20,27,29). Studies have demonstrated that such CWI can rapidly lower core and skin temperatures to preexercise levels and below for up to 90 min postimmersion (20,28). It remains unknown whether these decreases in body temperatures persist after 90 min and whether any such decrease could affect subsequent sleep. For example, it is possible that CWI performed after evening exercise could accelerate the normal presleep rate of decline of core temperature and therefore augment sleep propensity.

The purpose of this study was to use laboratory-based polysomnography (PSG) to determine whether high-intensity evening exercise, either followed by CWI or not, influenced subsequent sleep in healthy young male athletes.

Back to Top | Article Outline



Fifteen male cyclists and triathletes, cycling at least 150 km per week, were initially recruited for this study. They were in an endurance training phase, leading up to competitions within the next few months. The study design required participants to complete two familiarization sessions and three experimental sessions for 5 wk. Subsequently, two participants withdrew during the study due to illness and two were excluded, as they were diagnosed with mild obstructive sleep apnea (apnea–hypopnea index = 5.9 and 9.4 events per hour of sleep) after the sleep familiarization night.

A total of 11 participants completed all tests (mean ± SD; age = 26.0 ± 4.4 yr, body mass = 73.9 ± 7.3 kg, cycling peak power = 409.1 ± 43.7 W, V˙O2max = 65.5 ± 8.3 mL·kg−1·min−1, sum of seven skinfolds = 50.2 ± 10.5 mm). Study procedures were approved by The University of Western Australia’s Human Research Ethics Committee, and all participants provided written informed consent.

Back to Top | Article Outline

Familiarization sessions.

The first familiarization session was undertaken within 2 wk of the experimental trials and involved a graded cycle ergometer exercise test (Evolution cycles, Geelong, Australia) to assess peak power, V˙O2max, and lactate threshold as previously described (23) and was performed within 2 wk of the experimental trials. The second familiarization session was held on a separate evening and required participants to undergo a full laboratory-based overnight sleep study using PSG. Electrodes were attached according to the American Academy of Sleep Medicine recommendations (2): six electroencephalogram electrodes, two electrooculogram channels, five EMG channels (three chin and two legs), oxygen saturation via pulse oximetry, ECG, thoracic and abdominal respiratory effort via belts, oral/nasal thermistor, nasal prongs, and position sensor. Studies were analyzed according to standard guidelines (10). During this session, participants were also fitted with a wrist actigraph (Micro Mini-Motion Logger; Ambulatory Monitoring Inc., Ardsley, NY), which they were asked to wear for the duration of the study (4 wk). Participants were also required to complete sleep and physical activity diaries throughout this time.

Back to Top | Article Outline

Experimental testing sessions.

On three separate occasions, participants were randomly assigned to one of three different experimental conditions: CWI postexercise, exercise alone (EX) or control (CON: no exercise or CWI). Each experimental session was initiated on the same day of the week, at the same time, for three consecutive weeks. Seven participants commenced testing at 1830 h, two commenced testing 40 min later (1910 h), and another two commenced testing 60 min later (1930 h). Because of these slight differences in start time, the clock times mentioned in the remainder of the article (for circadian rhythm interpretation) reflect the start time of most participants (1830 h). Participants completed a food and fluid diary for the 12 h before each session, were required to maintain a similar daytime exercise regimen on the day before and the day of each test, and abstained from caffeine, alcohol, sports drinks, bananas, chocolate, and medications containing ibuprofen (as these products can interfere with the analysis of salivary melatonin) for 24 h before each session.

At home, 6 h before testing (∼1230 h), participants ingested a radiotelemetry pill (HQ Inc., Palmetto, Florida) for core temperature measurement. On arrival at the laboratory, participants were weighed (± 10 g; D-7470; August Sauter GmbH, Ebingen, Germany) then sat quietly while skin temperature sensors (iButtons; Embedded Data Systems, Lawrenceburg, Kentucky) were attached at four sites (sternum, forearm, thigh, and calf) to create one weighted overall measurement as reported by Ramanathan (21). Core temperature was then recorded and a heart rate monitor attached (Polar Electro Oy, Kempele, Finland). Subjective estimates of fatigue and recovery status were recorded using five-point scales (“How fatigued do you feel?”, where 1 = not fatigued at all, and 5 = very fatigued and “How recovered do you feel?”, where 1 = not recovered at all and 5 = very recovered) (8,24). After remaining seated for 15 min, baseline saliva samples were collected for melatonin analysis. Here, ambient light intensity was approximately 175 lx, as measured by a commercial light meter (Gossen Profisix, Erlangen, Germany). Core temperature and heart rate were recorded every 5 min during the exercise protocol, whereas skin temperatures were recorded only during the initial baseline period.

Participants performed intense cycling exercise (CWI and EX conditions) to produce fatigue similar to that experienced in competition/training (10 min controlled warm-up then 15 min at 75% peak power followed by a 5-min break then a 15-min maximal time trial) (23,27). In the CON trial, participants sat quietly in the same room (21.6°C ± 0.6°C, 57.0% ± 8.9% relative humidity) for 40 min. Fatigue and recovery status were measured immediately on completion of cycling or sitting. After these measurements, participants were allowed 15 min to relocate and change into their swimming costumes. For the CWI condition, participants were immersed (seated) to the midsternal level (arms out) in an inflatable water bath for 15 min with temperature maintained at 14°C (iCool Portacovery, Queensland, Australia). In the two experimental sessions not requiring CWI (i.e., EX and CON), participants sat next to the water bath for 15 min so that environmental conditions (lighting, temperature, and humidity) were constant on all testing occasions. Immediately after this, participants walked approximately 200 m to the sleep laboratory (along a minimally lit path <100 lx) where, after sitting for 15 min, another saliva sample (60 min postexercise, 2100 h) was collected and fatigue and recovery status again recorded. Participants then had a 90-s shower (26°C), changed into their sleep wear, and ate a standardized meal (2350 kJ; CHO = 57%, CHON = 18%, fat = 25%). For the ensuing 1.5 h, participants sat quietly in a dimly lit room (∼44 lx), where they were set up for their sleep study. During this phase (end of exercise–bedtime), core temperature was measured every 5 min and heart rate before and after the 15-min recovery period and then every 30 min. Because the focus of these studies was measurement of sleep architecture (sleep stages), the following measures were not obtained on the experimental nights: oral/nasal thermistor, nasal flow, respiratory effort sensors, position monitor, and leg EMG channels.

Approximately 2 h postexercise (2200 h, 30 min before bedtime), participants were asked to sit quietly in their beds (in the same room for each participant for all sleep studies) for a further 15 min while PSG signal quality was assessed and biocalibrations were performed. At 2.5 h postexercise (2230 h), fatigue and recovery status were measured, a saliva sample was obtained, and the lights were turned off. During this phase, core and skin temperatures were measured every 5 min and heart rate every 30 min. All participants were awakened 7.5 h after bedtime (10 h postexercise, 0600 h). On waking, measurements were obtained of salivary melatonin (after sitting upright for 15 min), core and skin temperatures, and fatigue and recovery status.

Back to Top | Article Outline

Melatonin analysis.

Saliva was collected in Salivette tubes (Serial No. 51.1534; Sarstedt, Nümbrecht, Germany), with plain cotton swabs placed into the mouth and moved around until heavy (∼5 min), then placed back into the Salivette tube. Samples were then centrifuged and the cotton plug discarded and frozen for later analysis of melatonin levels. Melatonin samples (200 μL) were sent to an independent Research Assay Facility (Adelaide Research Assay Facility, the University of Adelaide, South Australia) for analysis by radioimmunoassay using kits (Buhlmann Laboratories, Allschwil, Switzerland) following the manufacturer’s instructions. These kits use the Kennaway G280 antibody (30,31). The intra-assay coefficient of variation for melatonin concentration was less than 10%.

Back to Top | Article Outline


Overnight PSG was performed using recommended electrode and sensor placements (10) connected to the Grael PSG system (Compumedics Ltd., Melbourne, Australia) and analyzed using Profusion software version 3.4 (Compumedics Ltd.). The sleep familiarization night studies were scored to rule out any sleep disorders while the experimental nights were scored for sleep staging only (10).

The same experienced sleep technologist scored all experimental nights and was blinded to the condition for any given night’s study. The following parameters were calculated: total time in bed, total sleep time, sleep onset latency (SOL; time from lights out to the first epoch of sleep), REM onset latency, sleep efficiency (SE) [= (sleep duration − wake time) / sleep duration) × 100], wake after sleep onset (WASO), and percentage of time spent in each of the sleep stages (N1, N2, N3, and REM) compared with total sleep time.

Back to Top | Article Outline


Participants’ weeknight actigraphy data were obtained from the home environment (Monday through Thursday nights). A total of 5.5 ± 1.9 nights were obtained in each participant. A single representative “home environment” measure was obtained for each participant by averaging data obtained from all nights. ActiGraph measurements were recorded using the “zero crossing mode” and analyzed using the ACT Millennium and Action W2 software (Ambulatory Monitoring Inc.), according to the methods described by Richmond et al. (22). Measurements were obtained for total sleep time, WASO, SOL, and SE. Actigraphy measures obtained on the laboratory nights were compared against the average home environment measures.

Back to Top | Article Outline

Statistical analysis.

Core temperature, skin temperature, heart rate, melatonin, and fatigue and recovery measures for the CWI, EX, and CON conditions were analyzed using a linear mixed model with fixed effects of time, condition, and their interaction. A random effect of subject and subject within condition was also fitted to account for the correlation structure. Sleep PSG data (SOL, WASO, total sleep time, percentage of night spent in NREM sleep, N1, N2, N3, and REM sleep and REM onset latency) were also analyzed using a linear mixed model, but with condition as a fixed effect and a random effect of subject also fitted to account for the correlation structure. Sleep PSG data were further separated to analyze the first 180 min after sleep onset (during the time at which CWI had the greatest effect on core temperature during sleep). Both analyses were conducted in SAS version 9.3 (SAS Institute, Inc., Cary, NC). Cycling data (total work and mean power in each phase) were compared using paired t-tests. All data are presented as mean ± SD. Significance was set at P ≤ 0.05.

Back to Top | Article Outline


Exercise Tests

Total work and mean power during the 15-min cycling at 75% of peak power were similar for the CWI condition (270.7 ± 34.3 kJ and 300.2 ± 37.0 W, respectively) and EX condition (272.9 ± 33.1 kJ and 303.2 ± 36.7 W, respectively). During the 15-min time trial, total work was 4% greater (P = 0.01) and mean power 3% greater (P = 0.02) in the EX (260.9 ± 53.1 kJ and 291.0 ± 58.6 W, respectively) than CWI condition (249.4 ± 47.0 kJ and 278.0 ± 49.9 W, respectively).

Back to Top | Article Outline

Core Temperature

Significant differences (main effects) were noted for time (P < 0.001) and the interaction between condition and time (P < 0.001) (Fig. 1). There was no significant main effect for condition (P = 0.16). Compared with the CON condition, core temperature was higher immediately postexercise (2000 h) in the CWI and EX conditions (P < 0.001) and remained higher until 1 h postexercise when core temperature for the CWI condition was lower (∼0.3°C–0.4°C) than both EX and CON conditions (EX, P < 0.001; CON, P = 0.019). At 1.5 h postexercise (2130 h), core temperature in the CWI condition was lower than the EX condition only (P = 0.020) and at 2 h postexercise (2200 h) lower than the CON condition only (P = 0.011). Although core temperature was the same for all conditions at “lights off” (2.5 h postexercise, 2230 h), from 3.5 h (2330 h) until 4.5 h postexercise (0030 h), the CWI condition recorded a significantly lower core temperature (∼0.2°C–0.3°C) than both EX and CON (3.5 h, 2330 h CON P = 0.01, EX P = 0.003; 4 h, 0000 h CON P = 0.02, EX P = 0.002; 4.5 h, 0030 h CON P = 0.04, EX P = 0.03). Core temperature was similar for all conditions from 5 h (0100 h) postexercise until waking.



Back to Top | Article Outline

Skin Temperatures

Significant main effects were noted for time (P < 0.001) and the interaction between condition and time (P = 0.034) (Fig. 2). There was no significant main effect for condition (P = 0.72). At “lights off” (2.5 h postexercise, 2230 h), the mean skin temperature for the CWI condition was lower (∼0.7°C) than both the EX (P < 0.001) and CON (P = 0.003) conditions. At 3 h postexercise (2300 h), the mean skin temperature during the CWI condition was lower (∼0.5°C) than the EX condition (P = 0.02). Skin temperature was similar for all conditions from 3.5 h postexercise (2330 h) until waking.



Back to Top | Article Outline

Heart Rate

Significant main effects were noted for condition (P < 0.001), time (P < 0.001), and the interaction between condition and time (P < 0.001). Immediately postexercise, heart rate was higher in the CWI and EX conditions (∼175 bpm) compared with the CON condition (∼65 bpm) (P < 0.001) and remained greater compared with the CON condition until “lights off” at 2.5 h postexercise (2230 h). Heart rate was similar for all conditions from 3 h postexercise (2300 h) until waking.

Back to Top | Article Outline


A significant difference (main effect) was noted for time (P < 0.001) (Fig. 3). There was no significant main effect for condition (P = 0.96) or the interaction between condition and time (P = 0.84). In all conditions, melatonin levels increased from ∼4.6 pM at baseline to ∼23.5 pM at 1 h postexercise and to ∼40 pM at “lights out” (2.5 h postexercise, 2230 h). Melatonin levels on waking (10 h postexercise, 0600 h) remained elevated (∼23 pM) compared with baseline.



Back to Top | Article Outline

Perception of Fatigue

Significant main effects were noted for time (P < 0.001) and the interaction between condition and time (P < 0.001). There was no significant main effect for condition (P = 0.06). Immediately postexercise (2000 h), fatigue ratings for the CWI and EX conditions (both ∼4) were higher than the CON condition (∼2) (P < 0.001). At lights out (2230 h), fatigue scores were higher for the EX (∼3.5) compared with CON condition only (∼2.7) (P = 0.03). On waking (10 h postexercise, 0600 h,) fatigue scores were similar for all conditions (∼2.4).

Back to Top | Article Outline

Perception of Recovery

A significant main effect was noted for time (P < 0.001). There was no significant main effect for condition (P = 0.07) or the interaction between condition and time (P = 0.08). Recovery scores for the CWI and EX conditions decreased from approximately 3.5 at baseline (1855 h) to approximately 2.3 immediately postexercise (2000 h). By 1 h postexercise (2100 h), both conditions had returned to baseline levels (∼3.5) and were similar to CON scores. At lights out (2230 h), scores were similar between conditions and to baseline (∼3.6). On waking (10 h postexercise, 0600 h), recovery scores were similar for all conditions (∼3.5).

Back to Top | Article Outline


Whole night data.

There were no main effects between the three conditions for time in bed (P = 0.78), total sleep time (P = 0.47), SE (P = 0.35), SOL (P = 0.38), REM latency (P = 0.39), or the proportions of WASO (P = 0.21), N1 sleep (P = 0.89), N2 sleep (P = 0.21), N3 sleep (P = 0.76), NREM sleep (P = 0.09), or REM sleep (P = 0.08) (Table 1).



Back to Top | Article Outline

First 180-min PSG data.

The proportion of N2 sleep after the CWI condition was significantly higher than the CON condition (P < 0.001; Table 2). The proportion of NREM sleep was significantly higher after both the EX and CWI conditions than the CON condition (P = 0.009). A higher proportion of REM sleep was noted after the CON condition than the EX condition (P = 0.009) (Table 2). There were no differences (main effects) between conditions for total sleep time (P = 0.48), SE (P = 0.48), REM latency (P = 0.22), or the proportions of WASO (P = 0.48), N1 sleep (P = 0.22), or N3 sleep (P = 0.89).



Back to Top | Article Outline


There were no differences in any actigraphy-based measurements of sleep between the three experimental conditions (Table 3). SE was higher on the night of the CON condition (∼98%) than when at home (∼96%) (P = 0.04). Conversely, WASO was less (∼11 min difference) on the night of the CON condition than when at home (P = 0.03). SOL was shorter after both the CWI and the CON conditions (∼5–7 min) than at home (∼12.5 min) (P < 0.05). Total sleep time was not different when measured in the laboratory or at home (Table 3).



Back to Top | Article Outline


This study assessed the effect of evening exercise alone and evening exercise followed by CWI on subsequent sleep quality and quantity as compared to a control evening (without exercise or CWI) in young, healthy, competitive cyclists. When considering total overnight sleep, exercise alone or exercise and CWI did not influence any aspect of subsequent sleep quality or quantity. However, CWI decreased core body temperature (compared with EX and CON) by approximately 0.1°C–0.4°C for 5 h postexercise, which included the first 180 min of sleep. In addition, mean CWI skin temperature was lower (∼0.4°C–0.7°C) than EX and CON at bedtime and for the first 60 min of sleep. This initial period of sleep was characterized by more NREM and less REM sleep on nights after CWI than the control nights. Whether this could be attributable solely to CWI is unclear, as similar changes in sleep architecture were also observed in the initial sleep period after exercise without CWI.

The ecological validity of this study is high, as the intense cycling task used is similar to a hard training session cyclists would normally complete one to two times per week at a similar (evening) time. The study also had several novel aspects related to its experimental design. First, the use of actigraphy at home and during the laboratory-based sleep study allowed an examination of any differences between the laboratory-based sleep patterns and those when the athletes were at home. The similarity between these measures suggests that the sleep patterns observed in our laboratory studies were representative of sleep when at home. Second, we used the gold-standard measure of sleep, comprehensive laboratory-based PSG, to initially examine for the presence or absence of a sleep disorder and then to assess the effect of exercise and CWI on sleep. Third, we used a control condition (no exercise, no CWI) with which to compare any effects of exercise alone or exercise and CWI on subsequent sleep. A limitation of the data is that there was a small difference in the total work and mean power produced between the EX and the CWI conditions in the time trial. However, core temperature, heart rate, and fatigue ratings at the end of exercise were not different between CWI and EX.

Back to Top | Article Outline

Effects of evening exercise on sleep.

The approximately 1.5°C increase in core body temperature after high-intensity evening exercise observed in the present study is similar to that previously reported (23,28). When performed in the evening close (∼2–3 h) to bedtime, this increase in temperature could accelerate the normal rate of decline in core temperature, which is usually maximal (without prior exercise) around the time of sleep initiation (15). Although others have proposed that this could act to increase sleep propensity (34), our study did not support this hypothesis, as time to sleep onset was similar regardless of whether presleep core temperature was increased or not. Other measures of sleep quality and quantity were similarly unaffected by exercise-induced increases in core body temperature before sleep.

To our knowledge, only three previous studies have used PSG to examine the effect of evening exercise performed within 3 h of bedtime on sleep quantity and quality in healthy individuals (4,7,16). Myllymaki et al. (16) had young men and women undergo a graded cycling test to volitional exhaustion approximately 2 h before bedtime and reported an increase in the proportion of NREM sleep and a decrease in REM sleep. Possible reasons for these different findings compared with the present study may lie with the later time of exercise (2100 vs. 1900 h) used by Myllymaki et al. (16), their lower relative aerobic fitness levels (V˙O2max = 54 ± 8 mL·kg−1·min−1 vs. 65.5 ± 8.3 mL·kg−1·min−1), and their mixed gender sample, as core temperature fluctuations due to the menstrual cycle in females can affect resulting sleep (3). Furthermore, it is not clear from their study when during the night the increase in the proportion of NREM sleep and decrease in REM sleep occurred. Similar changes were observed in the present study, but only in the initial 3-h period of sleep (see following paragraphs).

Recently, Flausino et al. (7) reported improved sleep (increased SE, REM latency, decreased WASO, and proportion of N1 sleep) after moderate-intensity evening (∼2000 h) exercise in healthy, active men. However, comparisons with our study are again complicated, as their intensity of exercise was significantly lower (60 min at the first ventilatory threshold) than ours, which was reflected by a lower end-exercise core temperature (∼37.6°C compared with ∼38.7°C), and their participants were active (46.4 ± 7.2 mL·kg−1·min−1) rather than well trained as in our study. How greater (or lesser) aerobic fitness levels may impact on sleep is unknown at present. In contrast to these results, Bulckaert et al. (4) reported no significant differences in overall sleep architecture after 1 h of moderate-intensity exercise 2 h before bedtime, reporting only higher variance in heart rate during slow wave sleep in healthy participants.

Netzer et al. (18) did use well-trained cyclists and PSG to investigate the potential influence on catecholamines on sleep architecture. In comparison with a control night, they found a significant decrease in the proportion of REM sleep in the first half of the night, approximately 5 h after completing an afternoon race, which is similar to the results found here in the first 3 h of sleep. A significant correlation between REM latency and epinephrine excretion was reported (r = 0.63), but further research is required to assess the overall influence of catecholamine excretion during exercise on subsequent sleep. High-intensity training (similar to as performed here) will stress the autonomic nervous system, but in trained athletes, a rapid recovery is common (26) and does not appear to affect subsequent sleep quality (17).

Melatonin is a natural hormone released by the pineal gland and influenced by ambient light levels. Its production is tightly linked to the sleep–wake cycle, with levels being elevated before and during sleep but barely detectable during the day. This circadian rhythm of melatonin secretion is inversely proportional to the normal core temperature cycle, with increases in melatonin promoting decreases in core temperature via cutaneous dilation (32). Despite the differences in temperature between the exercise and the control conditions, there was no difference in the melatonin levels in any condition at any time points. As melatonin is suppressed by the presence of bright light (6), care was taken here to minimize any effect of light as much as practically possible, with lighting levels restricted to approximately 44–175 lx only before bedtime. This was deemed to be effective as the 1-h postexercise (2100 h) and bedtime (2230 h) levels were significantly above baseline values.

There were no differences in the perception of recovery between any condition. As expected, fatigue scores increased postexercise in the EX and CWI conditions compared with the CON, and heart rate remained elevated above CON in both conditions until bedtime (2.5 h postexercise, 2230 h). Interestingly, fatigue scores were higher at sleep onset in the EX condition compared with the control (3.5 vs 2.7); however, next morning at wake, all conditions were the same (∼2.4) as were heart rate values. These results underpin the importance of sleep as a postexercise recovery tool (25).

Back to Top | Article Outline

Effect of evening exercise and CWI on sleep.

Commonly, CWI is used as a postexercise recovery technique, with athletes routinely immersing themselves in water of approximately 14°C for approximately 15 min after training or competition in the evening (20,27,29). We hypothesized that CWI performed postexercise may potentially further increase sleep propensity (over exercise alone), as CWI performed after the same exercise protocol as used here can accelerate the decline in core temperature, achieving lower levels than preexercise for up to 90 min postimmersion (23). The results from the current study are the first to show that CWI produced a slightly greater rate of decline in core temperature (∼2.2°C) beyond that of EX (∼2.0°C) until 2.5 h postexercise. In addition, after “lights out,” there was a 90-min period where core temperature decreased (∼0.3°C) beyond that of both the EX and CON conditions (3.5–4.5 h postexercise, 2330–0030 h). The CWI also produced slightly lower skin temperatures at bedtime and for the first 60 min of sleep than that in EX and CON conditions. Whether these small differences in core and skin temperatures produced by CWI are factors in the greater NREM (and lesser REM) sleep observed here in the first 180 min (compared with CON) is not clear, but the same finding was apparent with EX alone. Potentially, the rate of decline in core temperature (rather than the absolute level) and/or alterations in the core/shell ratio (12) may be influencing factors. In addition, these findings indicate that exercise rather than CWI may be driving any increased NREM and decreased REM sleep in the initial period of sleep. Such a finding is consistent with that of Myllymaki et al. (16), who also reported an increased proportion of NREM and decreased REM sleep postevening exercise but did not measure any core or skin temperatures.

The current study did not find any effect of CWI on melatonin levels. Kauppinen et al. (11) and Robey et al. (23) also found no change in melatonin levels for the initial 90 min after CWI postexercise. Our study is the first to measure melatonin levels after CWI postexercise, presleep, and postsleep, with results suggesting that early evening exercise and/or CWI does not affect melatonin, especially if light exposure is controlled.

Lastly, the CWI did not alter heart rate response postexercise in comparison with EX alone. However, CWI has also been known to immediately trigger postexercise parasympathetic reactivation (1). In respect to sleep, Harris (9) has stated that when transitioning from wake to NREM sleep, there is a decreased sympathetic and increased parasympathetic tone. Had this occurred in our study, we would have either seen a greater decrease in heart rate with CWI or changes in sleep quality in the CWI condition compared with either the CON or EX condition, but neither was observed in our study. Similarly, the perception of fatigue after CWI was not different to the EX condition either before sleep or after waking.

Back to Top | Article Outline


The results presented here are the first to show that both exercise alone and CWI performed after high-intensity evening exercise do not acutely affect overall night sleep quantity and quality compared with a control condition. Whether there are acute short-term effects in the first 3 h of sleep require further research for clarification. At present, coaching and sport science professionals can continue to prescribe high-intensity evening training sessions and use CWI postexercise in the knowledge that, while rapidly decreasing core and skin temperatures, CWI will not negatively affect subsequent whole night sleep.

The authors thank iCool sport for the loan of the chilling unit and inflatable bath.

No funding has been received for this work. The authors have no professional relationship with a for-profit organization that would benefit from this study. Publication does not constitute endorsement by the American College of Sports Medicine.

Back to Top | Article Outline


1. Al Haddad H, Laursen PB, Chollet D, Lemaitre F, Ahmaidi S, Bucheit M. Effect of cold or thermoneutral water immersion on post-exercise heart rate recovery and heart rate variability indices. Auton Neuro. 2010; 156: 111–6.
2. Ancoli-Israel S, Cole R, Alessi C, Chambers M, Moorcroft W, Pollak CP. The role of actigraphy in the study of sleep and circadian rhythms. Sleep. 2003; 26 (3): 342–92.
3. Baker FCF, Driver HS. Circadian rhythms, sleep, and the menstrual cycle. Sleep Med. 2007; 8 (6): 613–22.
4. Bulckaert A, Exadaktylos V, Haex B, De Valck E, Verbraecken J, Berckmans D. Elevated variance in heart rate during slow-wave sleep after late-night physical activity. Chronobiol Int. 2011; 28 (3): 282–4.
5. Buxton OM, Lee CW, L’Hermite-Baleriaux M, Turek FW, Van Cauter E. Exercise elicits phase shifts and acute alterations of melatonin that vary with circadian phase. Am J Physiol Regul Integr Comp Physiol. 2003; 284: R714–24.
6. Duffy JF, Czeisler CA. Effect of light on human circadian physiology. Sleep Med Clin. 2009; 4 (2): 165–77.
7. Flausino N, Prado J, Tufik S, De Mello M. Physical exercise performed before bedtime improves the sleep pattern of healthy young good sleepers. Psychophysiology. 2012; 49: 186–92.
8. Halson SL, Quod MJ, Martin DT, Gardner AS, Ebert TR, Laursen PB. Physiological responses to cold water immersion following cycling in the heat. Int J Sports Physiol Perform. 2008; 3: 331–46.
9. Harris C. Neurophysiology of sleep and wakefulness. Respir Care Clin N Am. 2005; 11 (4): 567–86.
10. Iber C, Ancoli-Israel S, Chesson A, Quan S. The AASM Manual for the Scoring of Sleep and Associated Events: Rules, Terminology and Technical Specifications. 1st ed. Westchester (IL): American Academy of Sleep Medicine; 2007. p. 23–47.
11. Kauppinen K, Pajari-Backas M, Violin P, Vakkuri O. Some endocrine responses to sauna, shower and ice water immersion. Arctic Med Res. 1989; 48 (3): 131–9.
12. Krauchi K. The human sleep–wake cycle reconsidered from a thermoregulatory point of view. Physiol Behav. 2007; 90: 236–45.
13. Krauchi K, Deboer T. The interrelationship between sleep regulation and thermoregulation. Front Biosci. 2010; 15: 604–25.
14. Morin CM, Hauri PJ, Espie CA, Spielman AJ, Buysse DJ, Bootzin RR. Nonpharmacologic treatment of chronic insomnia. An American Academy of Sleep Medicine review. Sleep. 1999; 22 (8): 1134–56.
15. Murphy PJ, Campbell SS. Night time drop in body temperature: a physiological trigger for sleep onset? Sleep. 1997; 20 (7): 505–11.
16. Myllymaki T, Kyrolainen H, Savolainen K, et al. Effects of vigorous late-night exercise on sleep quality and cardiac autonomic activity. Sleep Res. 2011; 20: 146–53.
17. Myllymaki T, Rusko H, Syvaoja H, Juuti T, Kinnunen M-L, Kyrolainen H. Effects of exercise intensity and duration on nocturnal heart rate variability and sleep quality. Eur J App Physiol. 2012; 112: 801–9.
18. Netzer NC, Kristo D, Steinle H, Lehmann M, Strohl KP. REM sleep and catecholamine excretion: a study in elite athletes. Eur J of Appl Physiol. 2001; 84: 521–6.
19. O’Connor PJ, Breus MJ, Youngstedt SA. Exercise-induced increase in core temperature does not disrupt a behavioural measure of sleep. Physiol Behav. 1998; 64 (3): 213–7.
20. Peiffer JJ, Abbiss CR, Watson G, Nosaka K, Laursen PB. Effect of cold-water immersion duration on body temperature and muscle function. J Sports Sci. 2009; 27 (10): 987–93.
21. Ramanthan NL. A new weighting system for mean surface temperature of the human body. J Appl Physiol. 1964; 19: 531–3.
22. Richmond L, Dawson B, Hillman DR, Eastwood PR. The effect of interstate travel on sleep patterns of elite Australian Rules footballers. J Sci Med Sport. 2004; 7 (2): 186–96.
23. Robey E, Dawson B, Halson S, Goodman C, Gregson W, Eastwood P. Post-exercise cold water immersion: effect on core temperature and melatonin responses. Eur J Appl Physiol. 2013; 113 (2): 305–11.
24. Rowsell GJ, Coutts AJ, Reaburn P, Hill-Haas S. Effects of cold-water immersion on physical performance between successive matches in high-performance junior male soccer players. J Sports Sci. 2009; 27 (6): 565–73.
25. Samuels C. Sleep, recovery, and performance: the new frontier in high performance athletics. Neurol Clin. 2008; 26: 169–80.
26. Seiler S, Haugen O, Kuffel E. Autonomic recovery after exercise in trained athletes: intensity and duration effects. Med Sci Sports Exerc. 2007; 39 (8): 1366–73.
27. Vaile J, Halson S, Gill N, Dawson B. Effect of cold water immersion on repeat cycling performance and thermoregulation in the heat. J Sports Sci. 2008; 26 (5): 431–40.
28. Vaile J, Halson S, Gill N, Dawson B. Effect of hydrotherapy on recovery from fatigue. Int J Sports Med. 2008; 29: 539–44.
29. Vaile J, Halson S, Gill N, Dawson B. Effect of hydrotherapy on the signs and symptoms of delayed onset muscle soreness. Eur J App Physiol. 2008; 102: 447–55.
30. Vaughan GM. New sensitive serum melatonin radioimmunoassay employing the Kennaway G280 antibody: Syrian hamster morning adrenergic response. J Pineal Res. 1993; 15: 88–103.
31. Voultsios A, Kennaway DJ, Dawson D. Salivary melatonin as a circadian phase marker: validation and comparison with plasma melatonin. J Biol Rhythms. 1997; 12: 457–66.
32. Waterhouse J, Drust B, Weinert D, et al. The circadian rhythm of core temperature: origin and some implications for exercise performance. Chronobiol Int. 2005; 22 (2): 207–25.
33. Weinert D, Waterhouse J. The circadian rhythm of core temperature: effects of physical activity and aging. Physiol Behav. 2007; 90: 246–56.
34. Youngstedt SD, Kripke DF, Elliott JA. Is sleep disturbed by vigorous late-night exercise? Med Sci Sports Exerc. 1999; 31 (6): 864–9.


© 2013 American College of Sports Medicine