A comprehensive search was conducted across the Scopus, Web of Science, PubMed, and Embase databases. The initial search was conducted on 23 April 2025, and was updated on 6 November 2025. Database searches used keywords combined with MeSH terms. Search terms included: “Stretch-Shortening Exercise” OR “Stretch Shortening Cycle” OR “plyometric training” OR plyometric OR plyometrics OR “jump training” OR “jump exercise” OR “ballistic training” OR “drop jump” OR “depth jump” AND “basketball” OR “soccer” OR “football” OR “handball” OR “volleyball” OR “rugby” OR “team sport”. The search was limited to titles and abstracts, with no restrictions applied to publication region, year, or language. We also searched PROSPERO and the Cochrane Database of Systematic Reviews for relevant protocols to determine whether they had been published.

Eligibility criteria were defined according to the PICOS framework and are summarized in Table 1. The age range followed the World Health Organization definition of adolescents (10–19 years) (Organization, 2023).

Duplicate references were identified and removed by one reviewer (FZ) using EndNote 21 (Clarivate Analytics, Philadelphia, Pennsylvania, USA). Two researchers (FZ and YL) then independently screened titles, abstracts, and full texts against the predefined criteria.

Basic information of the literature was extracted independently by one author (FZ), including: (1) author and publication year; (2) age and gender of the subjects; (3) sample size; (4) sport and athlete level; (5) intervention measures; (6) training duration, training frequency, and training volume; and (7) outcome indicators. The results were reviewed by a second author (YL). We first attempted to obtain missing or unclear data by directly emailing the corresponding authors. All discrepancies between reviewers were then resolved through discussion. For any persisting disagreements, a final decision was made by a designated senior reviewer (LS). To avoid overestimating the sample size, if a control group in a study is compared with multiple experimental groups, the sample size of the control group should be divided by the number of comparisons for allocation. Biological maturity information was extracted and summarized descriptively, with emphasis on the reported maturity metrics. Athlete level was reclassified in a standardized manner using the McKay Participant Classification Framework (PCF; Tier 0–5) (McKay et al., 2021). When multiple COD tests were reported in a study, only the longest test time was included in the analysis, defined as the COD protocol with the greatest total test distance or the highest number of directional changes.

Two assessors (FZ and JL) independently assessed risk of bias using the Cochrane RoB 2.0 tool. The assessment covered five domains: (D1) Randomisation process, (D2) deviations from intended interventions, (D3) missing outcome data, (D4) measurement of the outcome, and (D5) selection of the reported result. Judgements were made for each domain and overall, as low risk, some concerns, or high risk. The certainty of evidence for each primary outcome was assessed using the GRADE approach. As all included studies were randomized controlled trials, certainty started at high and was downgraded when applicable across the following domains: risk of bias, inconsistency, indirectness, imprecision, and other considerations (e.g., publication bias). The overall certainty for each outcome was rated as high, moderate, low, or very low (Guyatt et al., 2011). Disagreements were resolved by discussion, with arbitration by a third assessor when necessary.

The meta-analyses were conducted using Review Manager V.5.4.0 and Stata 15.0. A total of eight meta-analyses were performed: (1) CMJ, (2) CMJA, (3) SJ, (4) SLJ, (5) ≤10-m linear sprint, (6) 20-m linear sprint, (7) 30-m linear sprint, (8) COD. A meta-analysis was conducted when at least three independent studies reported the same outcome measure (Borenstein et al., 2021). The effect size (ES) is represented by Hedge’s g and calculated by the mean and standard deviation of each dependent variable before and after training. For time-based outcomes, negative effect sizes represent improvements in performance. Given the expected between-study differences in participants, training programmes, and testing protocols, a random-effects model was used to pool effect sizes. Pooled effects are presented as Hedges’g with 95% confidence intervals (Deeks et al., 2019). The effect size is explained by the following criteria: trivial (<0.2), small (0.2-0.6), moderate (>0.6-1.2), large (>1.2-2.0), very large (>2.0-4.0), and extremely large (>4.0) (Hopkins et al., 2009). Heterogeneity among studies was assessed using the I² statistic, categorized as low (<25%), moderate (25%–75%), or high (>75%) (Higgins and Thompson, 2002). Egger’s test was used to assess publication bias. When publication bias was detected, the trim-and-fill method was used (Duval and Tweedie, 2000). Sensitivity analysis was employed to ensure the robustness of the meta-analysis results. p < 0.05 was set as the threshold for statistical significance.

To explore potential sources of heterogeneity, subgroup analyses and meta-regression analyses were conducted. Age, training duration, and total number of jumps were taken as moderating variables. Specifically, age groups followed the WHO age-based developmental stage classification described in: 10–12.99 (pre-PHV), 13–15.99 (mid-PHV), and 16–18.99 years (post-PHV), which reflects chronological age rather than directly assessed maturity. The training duration and the total number of jumps lack standardized classifications. To ensure the subgroup analysis has sufficient statistical power, we grouped them according to the distributions observed in the included studies. Meta-regression analysis was conducted when at least 10 studies reported the same outcomes (Cumpston et al., 2019).

A preliminary literature search yielded 5,407 articles. After deleting duplicate literature, an initial screening was performed based on the title and abstract, followed by downloading and reading the full text. Finally, 31 studies met the inclusion criteria (Sankey et al., 2008; Sedano et al., 2011; Ozbar et al., 2014; Zribi et al., 2014; Attene et al., 2015; Hammami et al., 2016; Asadi et al., 2018; Hernández et al., 2018; Ramirez-Campillo et al., 2018; Fathi et al., 2019; Jlid et al., 2019; Meszler and Váczi, 2019; Ramirez-Campillo et al., 2019; Drouzas et al., 2020; Negra et al., 2020; Ramirez-Campillo et al., 2020a; Ramirez-Campillo et al., 2020b; Vera-Assaoka et al., 2020; de Villarreal et al., 2021; Noutsos et al., 2021; Padrón-Cabo et al., 2021; Palma-Muñoz et al., 2021; Paes et al., 2022; Gaamouri et al., 2023; Aztarain-Cardiel et al., 2024; BOGIATZIDIS et al., 2024; Haghighi et al., 2024; Liu et al., 2024; Sammoud et al., 2024; Türkarslan and Deliceoglu, 2024; Öztürk et al., 2025). The complete literature screening process is summarized in Figure 1.

Characteristics of study participants and intervention protocols are detailed in Table 2.

Detailed results of the bias risk assessment for each area and overall are presented in Figures 2, 3. The primary sources of risk of bias were “randomization process” and “deviation from the intended intervention”, as it is challenging to blind participants and assessors in sports training. Only six studies explicitly described the randomization process (Hernández et al., 2018; Fathi et al., 2019; Ramirez-Campillo et al., 2020a; Ramirez-Campillo et al., 2020b; Liu et al., 2024; Sammoud et al., 2024). Only two studies were rated as having a low risk of bias in the domain of deviations from the intended interventions (Attene et al., 2015; Palma-Muñoz et al., 2021). For the primary outcome, the GRADE evidence quality level is low or very low (see Table 3). Downgrading was mainly due to risk of bias, inconsistency, and imprecision, with suspected publication bias for several outcomes.

The results of the eight meta-analyses are presented in Table 4, and the corresponding forest plots are provided in Figures 4–11.

PT significantly improved jump performance (CMJ: ES = 0.89, 95% CI: 0.59–1.19, I² = 75%; CMJA: ES = 1.00, 95% CI: 0.14–1.86, I² = 74%; SJ: ES = 0.48, 95% CI: 0.22–0.75, I² = 48%; SLJ: ES = 1.10, 95% CI: 0.62–1.58, I² = 62%). For time-based outcomes, PT also improved linear sprint performance (≤10 m: ES = −0.59, 95% CI: −0.87 to −0.32, I² = 49%; 20 m: ES = −0.42, 95% CI: −0.63 to −0.21, I² = 33%; 30 m: ES = −0.97, 95% CI: −1.68 to −0.26, I² = 59%) and COD (ES = −0.73, 95% CI: −1.02 to −0.45, I² = 60%).

A total of 17 subgroup analyses were conducted (Table 5). For CMJ, significant between-subgroup differences were observed for age (p = 0.04) and training duration (p = 0.009), with the largest improvements in athletes aged 16–18.99 years and in programmes lasting ≥8 to <10 weeks. For SJ, significant between-subgroup differences were observed for training duration (p < 0.001), with the largest improvement in programmes lasting ≥8 to <10 weeks. Similarly, for the ≤10-m and 20-m linear sprints, training duration showed significant between-subgroup differences (p < 0.001), with the greatest improvements in programmes lasting ≥8 to <10 weeks. Detailed information on the pooled effect size, heterogeneity, and the number of included studies for each subgroup can be found in Table 5.

When there were at least 10 studies for the same outcome measure, age, training duration, and training volume (total number of jumps) were used as covariates in a meta-regression analysis (see Table 6). The results indicated that age was significantly associated with improvements in CMJ (β = 0.211, p = 0.026), ≤10-m linear sprint (β = −0.119, p = 0.031), 20-m linear sprint (β = −0.117, p = 0.023), and COD (β = −0.163, p = 0.048). Training duration was significantly associated with improvements in 20-m linear sprint time (β = −0.206, p = 0.034). The total number of jumps was significantly associated with improvements in the SJ, although the magnitude of the association was small (β = −0.00048, p = 0.049).

Publication bias was assessed only for outcomes with ≥10 studies. Therefore, Egger tests were performed for five outcomes (see Table 4). CMJ and COD showed a risk of publication bias. By using the trim-and-fill method for adjustment, the results' significance remained unchanged, indicating that publication bias did not significantly affect the effect size.

Sensitivity analyses indicated that the pooled ES was robust for CMJ, SJ, SLJ, all sprint outcomes, and COD, whereas the pooled ES for CMJA was sensitive to omission of individual studies (Supplementary Figures S1–S8).

The meta-analysis showed that PT can effectively improve jump performance in youth team-sport athletes. Specifically, PT significantly improved CMJ (ES = 0.89), CMJA (ES = 1.00), and SLJ (ES = 1.10), whereas the improvement in SJ (ES = 0.48) was smaller but still statistically significant. CMJA results should be interpreted cautiously because they were not robust in sensitivity analyses.

Notably, the improvement in CMJ was clearly larger than that in SJ, suggesting that PT may be more sensitive for jumps involving a countermovement. This may be related to the relatively long pause at the bottom position of the SJ (three to five s). Such a pause may reduce the use of elastic energy stored during the eccentric phase, forcing the movement to rely mainly on concentric contraction, and thereby limiting the contribution of the SSC (MacDougall and Sale, 2014; Stojanović et al., 2017). Improvements in jump performance following PT may be related to structural and neuromuscular adaptations, such as muscle fiber hypertrophy and improved tendon collagen properties, which increase tendon stiffness (Pääsuke et al., 2001; Shepstone et al., 2005), enhance the rapid recruitment of high-threshold motor units, improve central nervous system excitability and reflex control, and strengthen intermuscular and intramuscular coordination (Markovic and Mikulic, 2010; Seiberl et al., 2021).

Regarding potential moderators, subgroup and meta-regression analyses suggested that age and training duration may be associated with the CMJ training effect. The subgroup aged 16–18.99 years showed a greater improvement in CMJ performance (ES = 1.83, p = 0.04), and meta-regression similarly demonstrated a significant association between age and CMJ improvement (β = 0.211, p = 0.026). Importantly, the subgroup and meta-regression results were consistent, suggesting that age may be associated with PT responsiveness. A reasonable explanation is that in older adolescent athletes, a more mature central nervous system and higher levels of testosterone and growth hormone may promote structural and neuromuscular adaptations (Moran J. J. et al., 2017; Radnor et al., 2018; Ramirez-Campillo et al., 2023). In contrast, in younger athletes, lower hormone levels may limit structural adaptations, and their improvements are more related to neuromuscular optimization (Tumkur Anil Kumar et al., 2021). However, because age was used as a proxy rather than directly assessed biological maturity, this interpretation should be considered cautiously.

For training duration, CMJ showed a larger improvement in programmes lasting ≥8 to <10 weeks (ES = 1.92, p = 0.009), and the SJ duration subgroup showed a similar trend (ES = 1.08, p < 0.001). Taken together, the current evidence indicates that programmes lasting ≥8 to <10 weeks may yield clearer improvements in jump performance, without showing that “longer is always better.” The meta-regression indicated a very small negative correlation between the total number of jumps and SJ improvement (β = −0.00048, p = 0.049), which has limited practical significance and should be interpreted with caution. However, the current evidence is insufficient to support the idea that increasing the total number of jumps leads to better training outcomes. One possible explanation is that, because the neuromuscular system of adolescents is still developing, excessive training volume may lead to central nervous system fatigue and high energy expenditure, resulting in impaired neural regulation and the accumulation of metabolic stress. These factors may compromise explosive performance and increase the risk of sports-related injuries (Cairns, 2006; Schoenfeld, 2010; Enoka and Duchateau, 2016). In contrast, appropriately prescribed training duration and volume may facilitate the restoration of energy reserves and muscle tissue repair following high-intensity training, thereby promoting supercompensation (Şahin et al., 2022; Luo et al., 2025). Notably, rapid growth around peak height velocity (PHV) may be associated with a further increase in injury risk (Faigenbaum et al., 2009; Luke et al., 2011). Therefore, greater emphasis should be placed on balancing training load and recovery during this stage.

The meta-analysis showed that plyometric training (PT) can significantly improve linear sprint performance in adolescent team-sport athletes. PT significantly improved ≤10-m (ES = −0.59), 20-m (ES = −0.42), and 30-m sprint performance (ES = −0.97). In practical terms, adding PT to regular sport-specific training may benefit both short-distance acceleration and longer-distance sprint performance. PT may enhance sprint performance through improved neural drive and neuromuscular coordination, increased lower-limb stiffness, and improved rapid force production, which together may reduce ground contact time and increase step frequency (Ross et al., 2001; Mackala and Fostiak, 2015; Tomalka et al., 2020).

Regarding potential moderators, subgroup analysis indicated that for ≤10-m sprint, programmes lasting ≥8 to <10 weeks produced larger improvements (ES = −1.35), with significant between-subgroup differences compared with ≤7 weeks and ≥10 weeks (p < 0.001). A similar pattern was also observed for 20 m sprint (ES = −1.35, p < 0.001). These findings suggest that a duration of ≥8 to <10 weeks may be more favorable for sprint improvements, but this should not be interpreted as a definitive “optimal” training duration. A reasonable explanation is that shorter programmes may provide insufficient accumulated stimulus, whereas longer programmes may involve excessive SSC loading, which may reduce tendon stiffness and make muscle fatigue and a neuromuscular adaptation plateau more likely, thereby compromising recovery (Komi, 2000; Nicol et al., 2006).

For age-related effects, meta-regression showed that age was associated with the training effect for both ≤10-m and 20-m sprints (≤10-m: β = −0.118, p = 0.031; 20-m: β = −0.117, p = 0.023), suggesting that older athletes may achieve larger improvements. With growth and development, progressive central nervous system maturation and hormonal changes may facilitate structural and neuromuscular adaptations (Lloyd and Oliver, 2013; Radnor et al., 2018; Silva et al., 2022); meanwhile, increases in lower-limb length may optimize the stride length–frequency combination and translate more effectively into sprint performance (Asadi et al., 2018). In addition, older athletes typically have longer trained experience, more stable neuromuscular control, and more consistent sprint technique, which may facilitate the translation of PT stimuli into sprint gains (Fort-Vanmeerhaeghe et al., 2016). In terms of the total number of jumps, the current evidence is insufficient to support the claim that simply increasing the total number of jumps can lead to better sprint training results. Further research is needed to verify this.

The meta-analysis demonstrated that plyometric training (PT) significantly improves change-of-direction (COD) performance in adolescent team-sport athletes, with a moderate overall effect size (ES = −0.73). Good COD is essential in team sports such as basketball, football, and handball, as it is closely related to offensive and defensive actions, the creation of scoring opportunities, and may be associated with reduced injury risk. These findings indicate that incorporating PT into regular sport-specific training may enhance COD and contribute to improved overall sport performance.

During COD tasks, athletes are required to rapidly transition between deceleration and re-acceleration, a process that fundamentally relies on eccentric–concentric muscle actions and the stretch–shortening cycle (SSC) of the lower-limb musculature (Sheppard and Young, 2006; Chaabene et al., 2018). PT imposes substantial inertial and eccentric braking loads during the deceleration phase, which may enhance eccentric strength and neural drive, improve inter- and intramuscular coordination, increase SSC efficiency, and enhance balance and joint stability (Markovic and Mikulic, 2010; Granacher et al., 2015; Chaabene et al., 2018; Jimenez-Iglesias et al., 2024). Collectively, these adaptations may facilitate faster deceleration control and more efficient re-acceleration, ultimately leading to improvements in COD performance.

Regarding potential moderators, age-based subgroup analyses did not reach statistical significance; however, meta-regression indicated that age may be associated with the COD training effect (β = −0.163, p = 0.048), suggesting that training responses may differ across age groups. With respect to training duration, no significant differences were observed between subgroups, although meta-regression revealed a borderline trend (β = −0.282, p = 0.050). In terms of the total number of jumps, the current evidence does not support achieving greater improvements in COD performance simply by increasing the total number of jumps.

Several limitations of this study should be considered when interpreting the results. The risk of bias assessment revealed that most included studies were assessed to have “some concerns” or a “high risk” of bias, primarily in the “randomization process” and “deviation from the intended interventions”, as it is challenging to blind participants and assessors in sports training. Furthermore, the GRADE quality of evidence indicated that the quality of evidence for the outcome indicators was mainly low to very low. These limitations may bias estimates of the impact of PT on the physical fitness of adolescent team athletes. Only five of the included studies focused on female adolescents, limiting the applicability of the findings to female adolescent sports teams. Therefore, more research on female adolescents is needed in the future. The information on biological maturity was insufficient and inconsistent (8 articles for PHV, seven articles for Tanner, and 16 articles did not report), and the types of maturity indicators were not uniform, which limited the further examination of the differences in the maturity stages. Although this study used the WHO age-based developmental stages for age grouping, this grouping does not represent the directly measured biological maturity and may mask the impact of true maturity differences on training adaptation. Finally, the information on ground contact time was insufficiently reported: most studies did not provide quantifiable ground contact time data or unified monitoring methods, and some only made qualitative descriptions such as “quick landing”. Safety reporting was a major limitation of the evidence base. As adverse events were rarely reported, safety outcomes could not be synthesized, and PT safety remains uncertain in this population.

PT is a feasible and effective training method for enhancing the jumping, sprinting and COD abilities of youth team sports athletes. A training program conducted twice a week for a duration of ≥8 weeks to <10 weeks can lead to more stable improvements in multiple physical performance indicators. In practical training, the focus of the training should not merely be on achieving a high total number of jumps, but should be adjusted individually based on the athlete’s developmental stage, training experience, recovery status and season workload.