We conducted a systematic literature search across four electronic databases (PubMed, Web of Science, Scopus, and SPORTDiscus). Studies published from inception until April 2025 were included. The search strategy involved using Boolean operators AND and OR with the following keywords: “ballistic training”, “power training”, “plyometric*”, “stretch-shortening cycle”, “jump training”, “jump exercise*”, “women”, “girl*” and “female*”. The results of the systematic literature search from the four databases were combined and duplicates were removed. After the removal of duplicates, two researchers (GL and RZ) screened the search results based on the inclusion criteria. Any discrepancies between the two authors were resolved by consensus with a third author (KW). In addition, we screened the reference lists of both previous meta-analyses and the articles that met the inclusion and exclusion criteria. The detailed search strategies for each individual database are provided in Table 1.

The studies were screened using the PICOS (Participants, Intervention, Comparators, Outcomes, and Study design) method (Liberati, 2009). Table 2 lists the inclusion/exclusion criteria. The additional inclusion criteria were as follows: (1) experimental trials published in peer-reviewed English-language journals.

Microsoft Excel (Microsoft Corp., Redmond, WA, United States) was used to extract the means and standard deviations of the dependent variables before and after the intervention from the included studies. The first author (GL) extracted physical fitness indicators as dependent variables, along with participant characteristics (sample size, sport, and years of practice) and intervention details (frequency, duration, number of sessions, and total ground contacts). The second author (RZ) verified the accuracy and completeness of the extracted data. Any discrepancies between the two authors were resolved by consensus with a third author (KW). The age range for adolescents was defined according to the World Health Organization (10–19 years) (Organization, 2023), and the classification of adolescent and adult female athletes was based on the mean age reported in each included study.

The Physiotherapy Evidence Database (PEDro) scale was used to assess the methodological quality of the included studies (Maher et al., 2003). The quality assessment was interpreted using the following 10-point scale: a score of ≤3 was considered to indicate poor quality, 4-5 indicated fair quality, and 6–10 indicated high quality. The PEDro scale consists of 11 items designed to evaluate methodological quality. Each satisfied item contributes 1 point to the overall PEDro score (range 0–10 points). Item 1 was not included in the quality rating of the studies as it pertains to external validity. The methodological quality of each included study was assessed independently by two authors (GL and RZ), and any discrepancies between the two authors were resolved via consensus with a third author (KW).

The risk of bias was assessed at the study level using the latest version of the Cochrane risk-of-bias tool for randomized trials (ROB2) (Flemyng et al., 2023) from five domains: randomization process, deviations from intended interventions, missing outcome data, measurement of the outcome, and selection of the reported result. Each item was rated as low risk, high risk, or some concerns. The quality of each included study was independently assessed by two reviewers (GL and RZ), and discrepancies were resolved by consultation with a third reviewer (KW).

Data analyses were conducted using the meta and metafor packages in R version 4.4.3 (R Project for Statistical Computing). The pre- and post-training means and standard deviations for each dependent variable, including countermovement jump height, 20-m linear sprint performance, and change-of-direction performance, were used to calculate the effect sizes (ES; Hedge’s g) for each physical fitness indicator in both the PT and control groups. A random-effects model using the DerSimonian–Laird method was employed to account for variability between studies that might affect the PT effects (Deeks et al., 2019; Kontopantelis et al., 2013). ES values were expressed with 95% confidence intervals (95% CI). The calculated ES were interpreted using the following scale: 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 was assessed using the I² statistic, with values of <25%, 25%–75%, and >75% representing low, moderate, and high levels of heterogeneity, respectively (Higgins and Thompson, 2002). The risk of publication bias was explored for continuous variables (≥10 studies per outcome) using the extended Egger’s test (Egger et al., 1997). In cases of bias, the trim and fill method was applied for adjustments (Duval and Tweedie, 2000). Statistical significance was set at p ≤ 0.05.

Subgroup analyses were performed using median split techniques to divide moderator variables (frequency, training duration, total sessions and total ground contacts). The median was calculated when at least three studies provided data for the moderator variable.

We performed sensitivity analyses to assess the robustness of the summary estimates (e.g., p value, ES, I²). To examine the effects of each result from each study on the overall findings, results were analyzed with each study deleted from the model (automated leave-one-out analysis).

Two authors (GL and RZ) rated the certainty of evidence (i.e., high; moderate; low; very low) using the Grading of Recommendations, Assessment, Development and Evaluation (GRADE) (Guyatt et al., 2011; Zhang et al., 2019a; Zhang et al., 2019b). The evidence started at a high level of certainty (per outcome), but was downgraded based on the following criteria: (i) Risk of bias in studies: judgments were downgraded by one level if the median PEDro scores were moderate (<6) or by two levels if they were poor (<4); (ii) Inconsistency: judgments were downgraded by one level when I² was high (>75%); (iii) Indirectness: low risk of indirectness was attributed by default due to the specificity of populations, interventions, comparators and outcomes being guaranteed by the eligibility criteria; (iv) Imprecision: one level of downgrading occurred whenever < 800 participants were available for a comparison (Carr et al., 2008) and/or if there was no clear direction of the effects; When both were observed, certainty was downgraded by two levels. (v) Risk of publication bias: downgraded by one level if there was suspected publication bias.

The search process identified 4096 studies (504 from PubMed, 1814 from Scopus, 1076 from Web of Science, and 702 from SPORTDiscus), resulting in a total of 2172 studies after removing duplicates. Ultimately, 20 studies were included in this meta-analysis. Figure 1 illustrates the study selection process. Table 3 displays the characteristics of participants in the included studies.

Using the PEDro checklist, 4 studies were considered to be of moderate quality (4-5 points), and the remaining 16 studies were considered to be of high quality (6–10 points). The results of the methodological quality assessment are presented in Table 4.

Additionally, to provide a more comprehensive assessment of bias risk, we employed the Cochrane ROB2 tool to evaluate each domain of bias in the included studies systematically. The results of the ROB2 assessment are shown in Figure 2 and Figure 3. Overall, three studies were classified as having a low risk of bias, while the remaining seventeen were classified as raising some concerns. Regarding the randomization process, only five studies reported allocation concealment, with the others providing no relevant information; therefore, most studies in this domain were rated as “some concerns.” In terms of deviations from intended interventions, one study was rated as “some concerns” due to intervention adjustments. For missing outcome data, two studies were rated as “some concerns” because they excluded participants with a low completion rate. In other domains, all studies were rated as low risk.

The overall effects of PT on physical fitness are shown in Table 5. The forest plots are shown in Figures 4–6.

The results of the subgroup analyses are presented in Table 6.

For improvements in CMJ height among adolescent athletes, significantly greater gains were observed with ≥9 weeks of PT compared to ≤8 weeks [ES = 1.46 vs. 0.42, P = 0.017]. Likewise, PT programs involving ≥16 total sessions produced more substantial improvements than those with ≤14 sessions [ES = 1.32 vs. 0.35, P = 0.018].

For improvements in COD performance among adolescent athletes, significantly greater gains were observed in subgroup analyses when PT lasted ≥9 weeks, included ≥18 total sessions, or involved ≥1260 total ground contacts, compared to training durations of ≤7 weeks, ≤14 sessions, or ≤1188 contacts, respectively [ES = −2.69 vs. −0.10, P < 0.001]. These three comparisons were based on the same dataset.

According to the GRADE assessment (Table 7), the certainty of evidence was considered moderate to low for the main analyses.

The meta-analysis indicated that PT significantly improved CMJ height in both adolescent and adult female athletes, with adolescents demonstrating greater gains, although the between-group difference was not statistically significant. Subgroup analyses further revealed that adolescent athletes experienced significantly greater improvements in jump performance when the training duration was ≥9 weeks or the total number of sessions was ≥16, compared to those with lower training accumulation.

The improvement in CMJ height may result from various neuromuscular adaptations, including enhanced neural drive of the agonist muscles, changes in the stiffness of the muscle–tendon unit, improvements in muscle architecture (e.g., increased muscle fiber cross-sectional area and fascicle length), better intermuscular coordination, and heightened stretch reflex excitability (Markovic and Mikulic, 2010). According to the principle of training specificity, the typical explosive movements involved in PT reinforce neuromuscular control patterns similar to those used in CMJ, particularly during the eccentric–concentric transition of the SSC. This repeated stimulation may enhance the efficiency of elastic energy storage and release, thereby facilitating improvements in jump performance (Stojanović et al., 2017).

Compared to adult females, adolescent female athletes exhibited greater improvements in jump performance following PT, which may be attributed to the synergistic effect of training adaptation and natural developmental processes. Lloyd et al. noted that the neuromuscular system during adolescence has not yet reached the stable state of adults and is in a phase of dynamic maturation (Lloyd et al., 2011). On one hand, structural components such as tendon stiffness and joint stiffness in adolescents gradually optimize with age; meanwhile, neuromodulatory capabilities including motor unit recruitment efficiency, muscle preactivation levels, and stretch reflex responses also show age-related improvements (Radnor et al., 2018). This natural development can promote the enhancement of SSC function even without specialized training (Malina et al., 2004; Radnor et al., 2018). On the other hand, the muscle activation strategy of adolescents is not yet fixed and is still in the transition stage from “reactive protective inhibition” to “performance-enhancing excitation” (Lambertz et al., 2003). Compared with adults with stable neuromuscular function, their systems are more sensitive to plyometric training stimuli. Training can more easily enhance the elastic energy utilization efficiency and reflexive muscle activation effect in SSC (Lloyd et al., 2011). This synergy between “the basic improvement of SSC brought by natural development” and “the neuromuscular adaptive improvements induced by training” amplifies the promoting effect of PT on SSC efficiency (Lloyd et al., 2011). In contrast, the neuromuscular function of adult females has reached a mature and stable state, lacking such a synergistic advantage, thus limiting the magnitude of improvement.

In addition, training variable analyses revealed that longer training duration and a higher number of sessions significantly enhanced CMJ performance among adolescents, while no significant differences were observed across training variables in adult females. This may reflect a heightened sensitivity of the adolescent neuromuscular system during a critical window of developmental plasticity. In contrast, adult athletes typically exhibit higher baseline jump performance, and their adaptation potential may be closer to saturation, thereby limiting the influence of training variable differences on performance outcomes (Stojanović et al., 2017).

The meta-analysis results showed that PT significantly improved 20-m sprint performance in adolescent female athletes, whereas no statistically significant improvements were observed in adult females.

The sprint-enhancing effects of PT in adolescents may be attributed to a combination of physiological and neuromuscular mechanisms. First, PT enhances lower-limb maximal strength and explosive power, which contributes to increased stride length and horizontal force output during the acceleration phase (Christou et al., 2006; Markovic and Mikulic, 2010). Second, plyometric exercises often involve high-frequency explosive jumps and rapid eccentric-to-concentric transitions, which exhibit strong movement specificity to sprinting. These characteristics help reduce ground contact time, thereby improving step frequency and overall sprint efficiency (Rimmer and Sleivert, 2000). From a neural perspective, PT may increase motor unit recruitment, enhance joint proprioception, and improve neuromuscular control, which facilitates better coordination of movement rhythm and posture during sprint acceleration (Asadi, 2013a). These mechanisms may work synergistically to enhance sprint performance, particularly in the initiation and acceleration phases.

In contrast, adult females did not show significant improvements in 20-m sprint performance following PT intervention. Among the two studies included, Cherni et al. reported a non-significant trend of improvement in the PT group (Cherni et al., 2020). The authors suggested that the plyometric exercises used in their protocol involved relatively long ground contact times, which may not effectively replicate the neuromuscular demands of short contact times required during sprinting, thus limiting the transferability of training adaptations (Cherni et al., 2020). Beyond the limitations of training design, individual developmental stage may also influence the magnitude of training adaptations. Adolescents are in a period of rapid growth, and the ongoing development of limb length and skeletal structure naturally promotes longer stride length and improved sprint efficiency (Asadi et al., 2018; Silva et al., 2022). In contrast, adult females have completed physical maturation, and their sprint performance may have a lower ceiling for adaptation compared to adolescents. Given the limited number of studies involving adult females, further research is needed to explore the responsiveness of this population to PT interventions targeting sprint performance.

Further analysis of training variables in adolescents revealed a trend toward greater improvement with higher levels of training accumulation, although between-group differences were not statistically significant. Zhou et al. found that, among adolescent basketball players, total jump count was strongly correlated with both sprint and COD performance (Zhou et al., 2024). This may be explained by the fact that higher repetition volumes result in more frequent activation of joint mechanoreceptors, thereby enhancing proprioception and motor control, ultimately improving sprint acceleration performance (Asadi, 2013b).

The meta-analysis revealed that PT significantly improved COD performance in both adolescent and adult female athletes, with greater improvements observed in adolescents, although the between-group difference was not statistically significant. Subgroup analyses further showed that significantly greater improvements in adolescents occurred when the training duration was ≥9 weeks, the total number of sessions was ≥18, and the total number of ground contacts was ≥1260, compared to those with lower training accumulation.

The improvements in COD ability following PT may primarily be attributed to neuromuscular adaptations, including increased motor unit recruitment and firing frequency (Aagaard et al., 2002; Radnor et al., 2018). Specifically, COD performance depends on rapid force development, eccentric control of the thigh musculature, and efficient coordination of the lower-limb extensors during the eccentric-to-concentric transition. PT has been shown to enhance these key capabilities (Aagaard et al., 2002; Miller et al., 2006). These neuromuscular adaptations not only increase lower-limb force output but also optimize rhythm and postural control during high-speed directional changes, thereby improving overall COD efficiency.

Compared to adult females, adolescent female athletes demonstrated greater improvements in COD performance following PT, which is consistent with the improvements observed in their CMJ performance. This similar pattern supports our hypothesis that the greater improvements seen in adolescents may be attributed to a synergistic interaction between training-induced adaptations and natural maturation. COD performance may be more influenced by motor control factors, such as skill execution and coordination, rather than by strength or power alone (Young et al., 2002). Adolescents tend to exhibit greater plasticity in intermuscular coordination, stretch reflex excitability, SSC utilization, and neural drive to the prime movers (Markovic and Mikulic, 2010), which may explain their enhanced gains in COD performance following PT.

It is important to note that although the meta-analysis showed a significant improvement in COD performance among adolescent athletes following PT, a high level of heterogeneity was observed (I² = 89.1%). Sensitivity analysis further revealed that the statistical significance disappeared when either the study by Gaamouri et al. or Hammami et al. was excluded. Notably, as shown by the subgroup analysis, these three studies comprised the entire high training accumulation group (training duration ≥9 weeks, ≥18 total sessions, and ≥1260 total ground contacts), and this subgroup exhibited a more pronounced improvement in COD (ES = −2.69 vs. −0.10). This finding implies that the significant improvements observed in adolescents may have been largely attributable to studies involving higher training accumulation. Since improvements in COD performance rely more heavily on neuromuscular adaptations, insufficient training volume may not generate adequate muscle engagement or neural activation (Ramírez-Campillo et al., 2013). Therefore, we recommend appropriately increasing training accumulation while ensuring a balance between training load and recovery, in order to enhance PT adaptations in adolescent athletes. Future research should further investigate the dose–response relationship between training volume and neuromuscular adaptations to develop more targeted intervention strategies.

A key limitation of this systematic review and meta-analysis is the insufficient methodological consideration for female athletes. As Elliott-Sale et al. noted, previous studies involving female participants have often failed to account for the menstrual cycle in their methodologies, which further complicates the derivation of evidence-based recommendations (Elliott-Sale et al., 2021). Research has shown that fluctuations in estrogen levels during the menstrual cycle may influence central nervous system fatigue, tendon and ligament strength, and muscle function, thereby leading to decreased sports performance or impaired adaptive responses to training (Emmonds et al., 2019). In the present study, one included study reported that all participants had regular menstrual cycles (Cherni et al., 2020), while another explicitly acknowledged that the menstrual cycle was not considered in its research process (Gaamouri et al., 2023). For the remaining studies, it was impossible to determine whether necessary methodological considerations (e.g., controlling for menstrual cycle phases) were implemented to address the known physiological differences between sexes. Therefore, we encourage future similar studies to explicitly incorporate menstrual cycle monitoring or control to reduce the interference of hormonal fluctuations on research results.

Additionally, athletes’ maturation stage may also modulate the effects of PT, a factor that could not be fully explored in this study. Existing research indicates that the responses of male adolescent athletes to PT vary by maturation stage (Asadi et al., 2017; Moran et al., 2017). However, due to the limited sample size of female adolescent athletes included in this study, we were unable to further stratify different maturation stages to analyze their differential effects on PT outcomes. Thus, we encourage future studies to further investigate the impact of different maturation stages on the adaptive mechanisms of PT in female adolescent athletes.