Effect of acute exercise on the dynamics of testosterone levels: a systematic review of randomized controlled trials

Design

This systematic review adhered to the Cochrane Handbook of Selection and the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) criteria (Page et al., 2021). The protocol is registered with PROSPERO under the registration number CRD420251007222.

Search strategy

A literature search was performed in five principal databases: PubMed, Web of Science, Cochrane, Scopus, and Ebsco, as of March 31, 2025. The search procedure utilized keywords and Medical Subject Headings (MESH): ((Exercises OR Post-Exercise OR Acute Exercise OR After a Game) AND (Testosterone) AND (Randomized Controlled Trial)). Furthermore, to improve the search outcomes, the reference lists of chosen research were manually examined to discover potentially qualifying articles. Two authors (G.L. and Q.T.) independently searched for and screened studies according to predetermined criteria during the month of April, and duplicate articles were removed using EndNote 20. In the event of a disagreement, a third author (S.W.) was consulted until a consensus was achieved.

Inclusion and exclusion criteria

The inclusion criteria comprised (1) a RCT design, (2) the presence of both an intervention and a control group, and (3) the measurement of testosterone as an end variable, (3) indexed in the Journal Citation Reports (JCR).

The exclusion criteria were: (1) review articles and conference proceedings; (2) research utilizing animal models; (3) publications with a significant risk of bias or without full-text availability; and (4) outcome measures that could not be transformed into mean and standard deviation.

The studies reviewed included only healthy individuals, and the research encompassed all age groups, males, and females.

Data extraction

Two separate researchers (Q.T. and L.G.) employed a standardized technique in Microsoft Excel 2019 (Microsoft Corp., Redmond, WA, USA) for data extraction (refer to Table 1). The subsequent fields were methodically recorded: (1) Study characteristics: first author, year of publication, and sample size; (2) Intervention information: kind, duration, and intensity; and (3) Participant characteristics: age, gender, exercise experience, and percentage of body fat (%). All numerical data were verified against the original sources to ensure precision. Any differences that emerged during this procedure were addressed through consultation with a third researcher (S.W.).

Quality assessment

Two researchers (G.L. and Q.T.) independently performed bias evaluations use the Cochrane Risk of Bias Tool for Randomized Controlled Trials (RoB-2) (Zhang et al., 2025). The evaluation encompassed six domains: randomization sequence creation, allocation concealment, blinding (of participants/staff and outcome assessors), measurement bias, inadequate outcome data, reporting bias, and additional potential biases. Rating differences were resolved through organized conversations with an external methodologist (S.W.). Each element was classified according to the official RoB-2 decision-making algorithm as “low risk,” “high risk,” or “unclear risk” (Li et al., 2023).

Search results

The preliminary database search revealed a total of 10,122 potentially pertinent articles. Following deduplication, 6,583 distinct records were preserved for evaluation. The review of titles and abstracts rejected 6,554 publications that failed to satisfy the inclusion criteria. A comprehensive evaluation of the remaining 29 articles led to the deletion of 14 studies for the following primary reasons: (1) nonrandomized study design (n = 6), (2) lack of outcome data (n = 4), and (3) treatments that were not independent campaigns (n = 4). In all, 15 articles (Bekris et al., 2022; Chatzinikolaou et al., 2010; Copeland, Consitt & Tremblay, 2002; Hackney, Premo & McMurray, 1995; Hough et al., 2011; Ispirlidis et al., 2008; Knatauskaitė et al., 2022; Kraemer et al., 2001; Lane & Hackney, 2015; McCaulley et al., 2009; Mirazizi, Rahmati-Ahmadabad & Fatolahi, 2022; Ronsen et al., 2001; Schumm et al., 2008; Shariat et al., 2014; Willoughby & Taylor, 2004) satisfied the inclusion criteria for the systematic review, as illustrated in the PRISMA flowchart (Fig. 1).

Characteristics of included studies

Table 1 summarizes the main characteristics of the included studies. This study comprised 16 RCTs, encompassing a total sample size of 267 participants (189 males (70.8%) and 78 females (29.2%)). The sample exhibited a broad age range (15–69 years old, with 249 cases under 40 years old, accounting for 93.3%; 18 aged 40 and above, 6.7%). According to the participant classification framework (McKay et al., 2022), 146 subjects (54.7%) had a background in systematic training (including trained and highly trained levels), 112 (41.9%) were recreationally active, and a small number were classified as sedentary (9, 3.4%). The literature was published over a 30-year duration from 1995 to 2025. Geographically, North America was predominant, comprising six contributions from the United States (Hackney, Premo & McMurray, 1995; Kraemer et al., 2001; Lane & Hackney, 2015; McCaulley et al., 2009; Schumm et al., 2008; Willoughby & Taylor, 2004) and one from Canada (Copeland, Consitt & Tremblay, 2002). Europe followed with three from Greece (Bekris et al., 2022; Chatzinikolaou et al., 2010; Ispirlidis et al., 2008) and one each from the United Kingdom (Hough et al., 2011), Norway (Ronsen et al., 2001), and Germany (Knatauskaitė et al., 2022). Asia included one contribution each from Iran (Mirazizi, Rahmati-Ahmadabad & Fatolahi, 2022), Malaysia (Shariat et al., 2014) and China (Xu & Guo, 2025). Aerobic activity (Bekris et al., 2022; Copeland, Consitt & Tremblay, 2002; Hackney, Premo & McMurray, 1995; Hough et al., 2011; Ispirlidis et al., 2008; Knatauskaitė et al., 2022; Lane & Hackney, 2015; Mirazizi, Rahmati-Ahmadabad & Fatolahi, 2022; Ronsen et al., 2001) (9, 56.3%) and resistance training (Chatzinikolaou et al., 2010; Copeland, Consitt & Tremblay, 2002; Hough et al., 2011; Kraemer et al., 2001; McCaulley et al., 2009; Schumm et al., 2008; Shariat et al., 2014; Willoughby & Taylor, 2004; Xu & Guo, 2025) (9, 56.3%) constituted the primary intervention modalities in exercise intervention protocols, while two research (Copeland, Consitt & Tremblay, 2002; Hough et al., 2011) established multimodal exercise groups. The duration of exercise varied significantly (15–90 min/session), with training intensity distributed as follows: high-intensity interval training constituted 56.3% (9/16) (Bekris et al., 2022; Copeland, Consitt & Tremblay, 2002; Hackney, Premo & McMurray, 1995; Hough et al., 2011; Ispirlidis et al., 2008; Knatauskaitė et al., 2022; Lane & Hackney, 2015; McCaulley et al., 2009; Shariat et al., 2014), while moderate-intensity continuous training comprised 62.5% (10/16) (Chatzinikolaou et al., 2010; Copeland, Consitt & Tremblay, 2002; Knatauskaitė et al., 2022; Lane & Hackney, 2015; McCaulley et al., 2009; Mirazizi, Rahmati-Ahmadabad & Fatolahi, 2022; Ronsen et al., 2001; Schumm et al., 2008; Willoughby & Taylor, 2004; Xu & Guo, 2025). Notably, four studies (Copeland, Consitt & Tremblay, 2002; Knatauskaitė et al., 2022; Lane & Hackney, 2015; McCaulley et al., 2009) established a multi-intensity control experimental group.

Furthermore, among the trials encompassed in this study, 12 employed serum samples (Bekris et al., 2022; Chatzinikolaou et al., 2010; Copeland, Consitt & Tremblay, 2002; Hackney, Premo & McMurray, 1995; Ispirlidis et al., 2008; Lane & Hackney, 2015; McCaulley et al., 2009; Mirazizi, Rahmati-Ahmadabad & Fatolahi, 2022; Ronsen et al., 2001; Schumm et al., 2008; Xu & Guo, 2025), five utilized saliva samples (Hough et al., 2011; Knatauskaitė et al., 2022; Kraemer et al., 1998; Lane & Hackney, 2015; Shariat et al., 2014), and a mere one employed plasma samples (Hough et al., 2011). One trial utilized both serum and saliva samples (Lane & Hackney, 2015), while another employed both plasma and saliva samples (Hough et al., 2011). With regard to the measured indicators, 14 studies assessed total testosterone (Bekris et al., 2022; Copeland, Consitt & Tremblay, 2002; Hackney, Premo & McMurray, 1995; Hough et al., 2011; Knatauskaitė et al., 2022; Lane & Hackney, 2015; McCaulley et al., 2009; Mirazizi, Rahmati-Ahmadabad & Fatolahi, 2022; Ronsen et al., 2001; Schumm et al., 2008; Shariat et al., 2014; Willoughby & Taylor, 2004; Xu & Guo, 2025), while the remaining two measured free testosterone (Chatzinikolaou et al., 2010; Ispirlidis et al., 2008). With regard to the alterations in testosterone levels subsequent to exercise intervention, a total of ten trials documented a substantial increase in testosterone levels following exercise (Bekris et al., 2022; Chatzinikolaou et al., 2010; Copeland, Consitt & Tremblay, 2002; Hackney, Premo & McMurray, 1995; Hough et al., 2011; Lane & Hackney, 2015; McCaulley et al., 2009; Mirazizi, Rahmati-Ahmadabad & Fatolahi, 2022; Schumm et al., 2008; Willoughby & Taylor, 2004), five trials noted no statistically significant difference (Ispirlidis et al., 2008; Knatauskaitė et al., 2022; Kraemer et al., 2001; Ronsen et al., 2001; Xu & Guo, 2025), and a single trial reported a significant decrease in testosterone levels after exercise (Shariat et al., 2014).

Risk of bias

Figure 2 illustrates the classification of the included studies into low, moderate, or high quality based on the following criteria: (1) a trial was considered to be of low quality if the risk of bias for randomization or allocation concealment was high, while the risk for other items was negligible; (2) a trial was considered to be of high quality if the risk of bias for randomization and allocation concealment was low while the risk of bias for all other items was low or unclear; and (3) a trial was considered to be of moderate quality if the trial did not meet the criteria for high or low risk (Li et al., 2023). Specifically, three trials had a low risk of bias (Bekris et al., 2022; Mirazizi, Rahmati-Ahmadabad & Fatolahi, 2022; Schumm et al., 2008) and 12 trials had a moderate risk (Chatzinikolaou et al., 2010; Copeland, Consitt & Tremblay, 2002; Hackney, Premo & McMurray, 1995; Hough et al., 2011; Ispirlidis et al., 2008; Knatauskaitė et al., 2022; Kraemer et al., 2001; Lane & Hackney, 2015; McCaulley et al., 2009; Ronsen et al., 2001; Shariat et al., 2014; Willoughby & Taylor, 2004).

Effect of exercise type on acute post-exercise testosterone levels

Various exercise type exhibit considerable variability in the modulation of testosterone levels, which may arise from the interplay between activity-specific physiological stress and endocrine feedback systems. Aerobic exercise influences the dynamic equilibrium of testosterone secretion via multifaceted mechanisms, while acute exercise can directly enhance the steroidogenic potential of testicular mesenchymal stromal cells through augmented blood supply (Moir et al., 2024) and sympathetic stimulation (Watt & Spriet, 2004). Testosterone recovery post-aerobic exercise is notably swift, typically reverting to baseline levels within 1 h (Hackney, Premo & McMurray, 1995), likely due to its comparatively low metabolic strain, prompt reestablishment of energy substrates, and the ephemeral nature of oxidative stress (Ammar et al., 2020).

The neuroendocrine modulation induced by resistance training via mechanical loading stimuli is distinctive. Microdamage in muscle fibers caused by external loads activates protein synthesis pathways (Grgic et al., 2020) and directly promotes testosterone release while enhancing myocyte sensitivity to anabolic signals (Griggs et al., 1986). While acute resistance exercise markedly elevates testosterone concentrations, its recovery kinetics exhibit prolonged fluctuation cycles, with certain studies noting recovery duration extending to hours or beyond from baseline levels (Tremblay, Copeland & Van Helder, 2004), potentially linked to modified myocyte membrane permeability induced by resistance training (Wernbom et al., 2012), enhanced glycogen depletion (Hokken et al., 2021), and an augmented response of the HPA axis (Athanasiou, Bogdanis & Mastorakos, 2023). The duration of HPA axis activation is extended. The variation in hormonal response unique to exercise types indicates that the immediate effects of different exercise type on the HPG axis, along with the outcomes of chronic adaptation, must be integrated when formulating training regimens.

Effect of exercise intensity on acute post-exercise testosterone levels

Overall, circulating total and free testosterone levels rise immediately post-exercise and revert to or fall below baseline within 30 min (Kraemer et al., 2006; Ratamess et al., 2005); however, these fluctuations are significantly affected by the selected exercise intensity. A relative intensity or total work output is necessary to elicit a testosterone response. The intensity of exercise influences the temporal dynamics of testosterone metabolism by variably regulating the neuroendocrine system. Moderate to high-intensity exercise can elevate peak blood testosterone at the conclusion of exercise by stimulating the HPG axis (Kumagai et al., 2018; Lane & Hackney, 2015), whereas excessive intensity may lead to a substantial decrease in recovery testosterone levels due to over-activation of the HPA axis (Hackney, Premo & McMurray, 1995; Shariat et al., 2014). When duration or repetitions remain constant, higher intensity exercise provokes a more substantial testosterone response. Raastad, Bjøro & Hallén (2000) exhibited a more pronounced acute response in testosterone during a high-intensity regimen relative to a moderate-intensity regimen. Meta-analytic evidence indicates that moderate- and high-intensity exercise significantly elevate testosterone levels in men (D’Andrea et al., 2020), while low-intensity exercise necessitates an extended duration to induce a notable increase in testosterone (Galbo et al., 1977), highlighting the crucial modulation of hormonal response by the interaction of intensity and duration.

When exercise intensity surpasses physiological thresholds, it induces an imbalance in endocrine homeostasis. Although male endurance athletes experience acute increases in testosterone and cortisol immediately following fatigue-inducing running, free testosterone remains below baseline levels during the subsequent 24-h recovery period (Daly et al., 2005). In strength training, total and free testosterone levels in men diminished during the initial two days following high-intensity deep squats (10 sets of 10 repetitions at 70% of one repetition maximum (1 RM) or 20 sets of one repetition at 100% of 1 RM) (Häkkinen & Pakarinen, 1993). A comparable pattern has been confirmed in international rugby players during post-competition surveillance, indicating that total testosterone levels required 36 h to revert to pre-competition values (Cunniffe et al., 2010). This may result from high-intensity exercise activating the HPA axis, which elevates cortisol levels and creates an antagonistic connection with testosterone. This hormonal imbalance may result in suboptimal testosterone levels by suppressing testosterone production (Choi, Chung & Lee, 2012). Consequently, accurate regulation of exercise intensity thresholds is crucial for sustaining hormonal equilibrium.

Effect of sample origin on acute post-exercise testosterone levels

In clinical practice, the identification of testosterone levels using biomarkers relies on two matrices, blood and saliva, which exhibit notable methodological discrepancies. In the study by Hough et al. (2011), Lane & Hackney (2015), simultaneous testing of serum and saliva samples revealed that acute exercise induces significant changes in serum and salivary testosterone levels. This change demonstrates a consistent pattern of elevation during both moderate-intensity and high-intensity exercise. Notably, under high-intensity exercise conditions, the response of salivary testosterone levels shows a stronger correlation with serum testosterone levels. Serum testing, the conventional gold standard (Fink, Bentzen & Horie, 2025), possesses a robust capacity to eliminate interference from other steroid hormones (Viette et al., 2011). The clinical application of this method is constrained by several factors: it necessitates the participation of healthcare professionals, and there are obstacles to the practicality of sampling for dynamic monitoring during exercise. Additionally, samples must be transported at -80°C and analyzed within 24 h, which presents a challenge for accessibility in resource-limited regions (Patjamontri et al., 2023).

In contrast, saliva testing has shown distinct benefits as a non-invasive option for youngsters, elderly individuals, and large-scale epidemiological screening (Horie, 2010; Xia et al., 2014). However, the precision and dependability of salivary tests are frequently influenced by many circumstances. Research indicates that while saliva testosterone is effective for detecting large doses of exogenous testosterone (Schönfelder et al., 2011), it poses challenges for accurately measuring endogenous testosterone levels. In twelve trained male subjects, low-intensity exercise resulted solely in an increase in serum total testosterone without notable alterations in salivary indices, while both exhibited a coordinated response during moderate- and high-intensity exercise (Lane & Hackney, 2015), indicating that salivary testing possesses greater ecological validity for acute hormonal fluctuations elicited by high-intensity exercise.

Effect of gender on acute post-exercise testosterone levels

Gender disparities markedly influence the kinetic profile of testosterone caused by exercise stress. Signals for gonadal testosterone synthesis and secretion originate in the hypothalamus, where specialized neurons synthesize and release gonadotropin-releasing hormone (GnRH). This hormone enters the anterior pituitary gland, stimulating the production of gonadotropins and the release of luteinizing hormone, which in turn promotes testosterone synthesis in testicular mesenchymal cells in males and in ovarian membrane cells in females (Kim, 2007). In females, testosterone is converted into oestradiol in granulosa cells neighbouring the membrane cells. Evidence indicates that a single session of high-intensity exercise causes immediate increases in serum total and free testosterone levels in male individuals, with peaks generally observed right after exercise and reverting to baseline or lower levels within 30 min (Copeland, Consitt & Tremblay, 2002; Kraemer et al., 1998; Lane & Hackney, 2015; Ronsen et al., 2001; Schumm et al., 2008; Willoughby & Taylor, 2004). It is notable that the magnitude of the hormonal response was influenced by the exercise regimen. The molecular mechanisms driving the exercise-induced testosterone response in women differ fundamentally from those in men. The primary site of testosterone production is the testicular interstitial cells, located solely in the testis. In women, androgen fluctuations may pertain to the dehydroepiandrosterone (DHEA) conversion pathway secreted by the reticular zona of the adrenal cortex, which largely accounts for the circulating testosterone levels in men, approximately ten times greater than those in women (Kvorning et al., 2007).

Limited research has directly investigated the impact of specific acute exercise on the post-exercise testosterone response in women. While some studies have documented significant increases in free or total testosterone following resistance training (Copeland, Consitt & Tremblay, 2002; Nindl et al., 2001), there remains considerable variability in findings across similar studies, with most reporting no acute elevation in post-exercise testosterone levels (Kraemer et al., 1993; Linnamo et al., 2005). Linnamo et al. (2005) established that under three distinct loading conditions: maximal (10 RM), submaximal (70% of 10 RM), and explosive (40% of 10 RM), there was no alteration in testosterone following five sets of 10 repetitions of sit-ups, bench press, and leg press. In males, identical maximal loading conditions led to a substantial elevation in testosterone (Linnamo et al., 2005). Kraemer et al. (1993) similarly shown that altering rest time, length, or load while maintaining a constant total volume during high-intensity exercise did not yield elevated post-exercise testosterone concentrations compared to baseline in women. In males, GnRH analogues suppress circulating luteinizing hormone, subsequently inhibiting testicular interstitial cell activity. This results in depot testosterone levels and obstructs acute resistance exercise-induced elevations in total and free testosterone. Consequently, testicular interstitial cells seem to be pivotal for the significant surge in post-exercise testosterone, potentially explaining the lack of testosterone responsiveness to resistance exercise in females (Vingren et al., 2010). Nonetheless, there is an insufficiency of robust, research-consistent evidence to substantiate this concept, and systematic validation via standardized experimental paradigms is urgently required.

Effect of age on acute post-exercise testosterone levels

The immediate effects of exercise on testosterone levels are significantly affected by age. According to Knatauskaitė’s research, acute exercise does not induce substantial alterations in testosterone levels among adolescent children, irrespective of gender (boys or girls) or the intensity of the exercise. Even post-puberty, when baseline testosterone levels begin to rise, no exercise-induced increase in testosterone is observed (Pullinen et al., 2002, 1998). Pullinen et al. (1998) established that following vigorous squatting exercises, plasma norepinephrine levels in teenage boys, along with total testosterone levels, were lower than those in adult males, and no association between these two responses was detected. testosterone levels in high school-aged subjects did not exhibit significant changes following the same workout regimen as that of college-aged subjects (Fahey et al., 1976). The restricted post-exercise testosterone response in adolescent children may result from the testes’ inability to swiftly augment testosterone secretion, insufficient metabolic stimulation during exercise, and the observation that adolescent athletes with extensive weightlifting training experience exhibit a more pronounced testosterone response than novice athletes during exercise (Kraemer et al., 1992).

The extent of exercise-induced alterations in testosterone levels appears to diminish with age following sexual maturity, particularly evident for free testosterone. The disparity between free and total testosterone levels may be attributed to elevated sex hormone-binding globulin (SHBG) concentrations and diminished albumin concentrations associated with ageing (Rodriguez et al., 2007). Acute post-exercise testosterone responses seem to be less significant in middle-aged and older adults (≥38 years) compared to younger adults (20–36 years) (Baker et al., 2006). Kraemer et al. (1998), when comparing acute post-exercise hormonal reactions in younger and older men, discovered that there were age-related changes in endocrine responses to exercise, with the most pronounced response being that of free testosterone. Furthermore, for young men, evening exercise may have a more pronounced effect on hormones compared to morning exercise (Xu & Guo, 2025). This indicates that the body can elicit physiological stimulation of the endocrine system via acute exercise, resulting in variations in testosterone levels, which are affected by age, as younger males exhibit a greater capacity for exercise-induced testosterone responses compared to older males. Additionally, the timing of physical activity can also have a significant impact on the body’s response.

Limitation

This study has several limitations. The included studies exhibited significant heterogeneity in exercise intervention protocols and participant characteristics, and testosterone measurement methods and metrics varied, collectively limiting the direct comparability of results. As only one study comparing exercise at different time points was identified, it was not discussed separately. In light of the methodological heterogeneity that may compromise the interpretability of quantitative meta-analysis results, a meta-analysis was not conducted. Instead, a qualitative synthesis approach was employed to systematically present the evidence landscape. Future research endeavors must entail the execution of larger, rigorously designed studies that employ standardized measurement criteria to validate the acute effects of specific exercise parameters on testosterone levels.