Impact of training volume settings between unilateral training and bilateral training on athletic performance: A systematic review and meta-analysis

Abstract

Background

The unilateral/bilateral dimension refers to whether an action is performed by one side of the body alone or involves both sides simultaneously. Unilateral training (UT) and bilateral training (BT) have been shown to enhance athletic performance. However, there are differences in training volume settings between unilateral and bilateral training, making it essential to understand their characteristics for optimizing training protocols and improving competitive level.

Objective

This systematic review aimed to: 1) explore the effects of training volume settings between unilateral training and bilateral training on improvements in athletic performance (muscle strength, jump performance, sprint time, and change of direction time), and 2) investigate the moderating effects on testing metrics, training frequency, and training protocol on athletic performance.

Methods

Searches were conducted in PubMed, Medline, Cochrane Library, EBSCOhost, and Web of Science (Core Collection) on June 5, 2024, and updated on April 17, 2025. Included studies were randomized controlled trials that investigated the effects of UT versus BT on athletic performance in athlete. Pooled effects for each outcome were summarized using Standardized Mean Difference [Hedges' g (g)] through a three-level meta-analysis model, and subgroup analyses were used to explore moderators. The certainty of evidence was assessed using the GRADE approach.

Results

A total of nine high-quality randomized controlled trials were included, involving 225 athletes. The results showed that there was no significant difference in improving athletic performance whether the training volume of UT and BT was the same (g = 0.20 [-1.56, 1.15], I²-2 = 77.8 %; I²-3 = 0.0 %, low GRADE) or when the UT volume was twice that of BT (−0.04 [-0.14, 0.06], I²-2 = 93.8 %; I²-3 = 0.0 %, moderate GRADE). When the training volumes of UT and BT were the same, the impact on athletic performance was not significantly moderated by test items, training frequency, or training protocols. Neither unilateral test metrics (g = −0.84) nor bilateral test metrics (g = −0.90) showed significant improvement. There was no significant difference between training twice a week (g = −0.16) and training three times a week (g = −0.19). Similarly, there was no significant difference between conducting plyometric training (g = −0.16) and another instance of plyometric training (g = −0.19). When the UT volume was twice that of BT, it might be significantly moderated by test items, but training frequency and training protocols were likely not significant moderators. Unilateral test metrics (g = −0.39) and bilateral test metrics (g = 0.64) both showed significant improvements. There was still no significant difference between training twice a week (g = −0.14) and training three times a week (g = 0.13). Furthermore, there was no significant difference between conducting plyometric training (g = −0.01) and another instance of plyometric training (g = −0.21).

Conclusion

The training volume settings between unilateral and bilateral training may not have a significant impact on athletic performance. The testing metrics might be the significant moderating factors, whereas training frequency and training protocol are likely not significant moderators.

Prospero registration

CRD42024545511.

Graphical abstract

1. Introduction

Optimizing athletic performance is a fundamental goal in sports,¹ and performance levels are closely related to the technical characteristics of specific sports. Both unilateral and bilateral movement patterns are involved in athletic movements such as sprinting, jumping, and throwing. The unilateral/bilateral dimension indicates whether an action is performed by one side of the body alone or involves both sides simultaneously.² In recent years, researchers have gained deeper insights into the respective contributions of unilateral and bilateral movements to performance enhancement. Based on these findings, researchers have validated the effects of targeted unilateral training (UT) and bilateral training (BT) on athletic performance. For example, UT is considered more effective in enhancing key techniques primarily relying on unilateral force.³ On the other hand, BT has been shown to improve strength,⁴ linear sprint speed over short distances,⁵ change of direction speed,⁶ and jump height.⁴ Therefore, understanding the characteristics of UT and BT and their impact on athletic performance is crucial for optimizing training protocols and elevating the competitive level.

Current research on the effects of UT and BT resistance training on athletic performance suffers from significant methodological and explore moderator limitations. Liao et al.² found that UT more effectively enhances unilateral jump performance, whereas BT shows greater improvements in bilateral strength. However, their study relied on traditional two-level analysis without accounting for moderating variables or focusing on specific populations such as athletes. Zhang et al.⁷ conducted subgroup analyses (e.g., intervention duration, frequency, and protocols) among athletes but still conducted two-level model, failing to exclude outliers (effect sizes >3.00) or conduct sensitivity analyses, which may have inflated effect size estimates.⁸ Moran et al.⁹ reported that both UT and BT enhance movement speed but neglected to explore moderating factors affecting horizontal movement performance in depth. These studies share critical limitations, including methodological shortcomings such as the absence of three-level modeling or rigorous outlier handling, insufficient examination of moderators, and high participant heterogeneity, which limits generalizability. Regarding training volume, existing studies typically follow two approaches: equal volume (UT: BT), where BT [e.g., 3 sets 6 repetitions at 75 % one-repetition maximum (1RM)] is matched by UT (3 sets × 3 repetitions per side), and two times UT: BT, where UT matches the total repetitions of BT (e.g., 3 sets × 6 repetitions per side). Notably, no study has systematically investigated the differential effects of these training volume between UT and BT, nor have subgroup analyses addressed this gap, constraining evidence-based optimization of training protocols.

To address existing research gaps, this three-level meta-analysis aims to: (1) compare the differential effects of UT and BT under varying training volume ratios (UT: BT vs. two times UT: BT) across athletic performance (muscle strength, jump performance, sprint, and change of direction); and (2) examine how these effects are moderated by testing metrics (unilateral vs. bilateral performance), training frequency, and training protocol. The findings will provide evidence-based recommendations for optimizing athletic performance prescription.

2. Methods

This review was performed by the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines.¹⁰ The completed PRISMA 2020 checklist is available in Electronic Supplementary Material Appendix S7. Additionally, this review has been registered in the PROSPERO database under the identifier CRD42024545511.

2.1. Information sources

Database searches were conducted in PubMed, Medline, Cochrane Library, EBSCOhost, and Web of Science (Core Collection). Eligible publications were full-text articles, with no restrictions on publication date, sample, or language; however, titles and abstracts had to be available in English. Three systematic snowballing searches were applied: 1) screening reference lists of included articles; 2) reviewing articles that cited the included articles; 3) exploring “similar articles” or “find similar” (PubMed). The searches covered records from inception to June 5, 2024, and was updated on April 17, 2025.

2.2. Search strategy

A Medical Subject Heading (MeSH) search was performed to identify all literature related to UT and BT, competitive sports, and athletic performance. Specifically, the database searches were performed using the keywords and truncations in conjunction with using the following search criteria: (unilateral OR bilateral OR Bilateral strength OR Unilateral strength OR Unilateral Strength Training OR Bilateral Strength Training OR Unilateral Training OR Bilateral Training) AND (muscle strength OR counter movement jump OR jump∗ OR sprint∗ OR change of direction) NOT ("animals” OR “disease” OR “disease").

2.3. Selection process

Deduplication of retrieved records was performed manually by an independent reviewer (M.L.T.) using Zotero [version 6.0.37]. Subsequently, the deduplicated records were then screened by two independent researchers (C.W.Z. and Y.O.S.), who assessed titles and abstracts based on predefined inclusion and exclusion criteria. Disagreements were resolved by a third researcher (M.L.T.). The same two researchers independently reviewed the full texts of eligible studies for final inclusion. Additional relevant articles were identified through reference lists of previous systematic reviews and expert recommendations from the research team, capturing studies potentially missed during the initial database search.

2.4. Eligibility criteria

This systematic review established clear inclusion and exclusion criteria based on the PICOS framework (Population, Intervention, Comparator, Outcome, Study Design). The inclusion criteria specified human athlete participants with systematic training experience. Animal studies were excluded during the title/abstract screening phase.

The intervention criterion involved training programs that included both UT and BT. The intervention had to have clearly defined protocol variables and last for a minimum of two weeks. Supervised interventions were considered eligible. The comparator criterion required studies to compare the effects of unilateral UT versus BT.

Eligible studies were required to report at least one quantitative outcome measure: muscle strength, jumping performance, sprinting performance, or change of direction ability. Studies are needed to perform statistical comparisons of pre- and post-intervention values relative to baseline/pre-training. Studies focusing on molecular-level physiological processes were excluded.

Regarding study design, only controlled between-group trials were included, specifically parallel randomized controlled trials or pre-post randomized crossover trials. The following types of studies were excluded: acute studies, review articles, opinion/viewpoint articles, validation studies, books, and case studies. Additionally, studies that did not include a comparison of UT and BT or only compared UT/BT with no-training groups were also excluded.

2.5. Data extraction and conversion

Data extraction was conducted by two reviewers (C.W.Z. and Y.O.S.), utilizing a customized extraction worksheet in Excel that was finalized prior to the full-text review. Reviewers independently extracted information on author details, study characteristics, participant demographics, training protocols, and outcomes. Discrepancies were resolved through discussion between the two reviewers, and no third researcher was required for arbitration. If data were missing or presented only in graphical form, the authors were contacted to request the necessary information. If this was unsuccessful and data remained in graphical form, relevant data were extracted using WebPlotDigitizer 4.1 (https://automeris.io/WebPlotDigitizer).¹¹ If studies for which missing data could not be obtained were excluded from the final analysis, however, this was not necessary. For each group, the mean, standard deviation (SD), and sample size were extracted pre- and post-intervention.

We extracted the mean, SD, and sample size reported for each group pre- and post-intervention. We pooled effects using pre- and post-intervention differences ($\mathrm{M}\pm \text{SD}$) for each outcome indicator. The mean difference (${\mathrm{M}}_{\text{change}}$) and SD of the change (${\text{SD}}_{\text{change}}$) were calculated using the following formulae,12, 13, 14 the first step involved calculating the difference in means:

$${\mathrm{M}}_{\text{change}}={\mathrm{M}}_{\text{post}}-{\mathrm{M}}_{\text{pre}}$$ (1)

where ${\mathrm{M}}_{\text{change}}$ is the raw mean difference, ${\mathrm{M}}_{\text{post}}$ is the reported mean post-intervention, and ${\mathrm{M}}_{\text{pre}}$ is the reported mean pre-intervention.¹⁵ Then the ${\text{SD}}_{\text{change}}$ is calculated as follows¹⁵:

$${\text{SD}}_{\text{change}}=\sqrt{{\text{SD}}_{\text{pre}}^2+{\text{SD}}_{\text{post}}^2-\left(2\times r\times {\text{SD}}_{\text{pre}}\times {\text{SD}}_{\text{post}}\right)}$$ (2)

where ${\text{SD}}_{\text{change}}$ is the SD of the difference in means, ${\text{SD}}_{\text{pre}}$ is the SD from pre-intervention, is the SD from post-intervention, and ${\text{SD}}_{\text{post}}$ is the correlation coefficient.¹⁵ Correlation coefficients for pre- and post-intervention were rarely reported in the included studies and were generally assumed to be r = 0.50, as suggested by the Cochrane Handbook.¹⁵ Referencing other studies and meta-analyses with similar outcomes for pre-and post-intervention Pearson's correlation coefficients (r), the obtained correlation coefficients were r = 0.84 for muscle strength,¹⁶ r = 0.87 for jumping performance,¹⁶ r = −0.77 for sprinting performance,¹⁷ and r = −0.89 for change of direction.¹⁸ While a simpler alternative is to use pooled SD of baseline scores, both and pooled SD of baseline scores are recommended,¹⁹ each with advantages depending on the research question.^14,20, 21, 22

Considering the relatively small sample sizes in most studies, Hedges' g was used as the mean effect size point estimate in each analysis, using the following formula²³:

$${\text{Hedges}}^{\prime }g=\frac{\left(BT\left[{\mathrm{M}}_{\text{change}}\right]-\text{UT}\left[{\mathrm{M}}_{\text{change}}\right]\right)}{{\text{SD}}_{\text{pooled}}}\times \left(1-\frac{3}{4\left({\mathrm{n}}_1+{\mathrm{n}}_2-2\right)-1}\right)$$ (3)

where ${\mathrm{M}}_{\text{change}}$ is the mean difference between the UT and BT groups, n₁ and n₂ are the sample sizes of these 2 groups, and ${\text{SD}}_{\text{pooled}}$ is the pooled SD of the measurements.²³ The specific formula is as follows:

$${SD}_{pooled}=\sqrt{\frac{\left(\left(n_1-1\right)\times {SD}_1^2+\left(n_2-1\right)\times {SD}_2^2\right)}{\left(n_1+n_2-2\right)}}$$ (4)

where $n_1$ and $n_2$ are the sample sizes of the 2 groups, and SD₁ and SD₂ are the SDs of both groups. Hedges' g were classified as trivial (<0.2), small (0.2–0.5), medium (>0.5–0.8), and large (>0.8).²⁴

2.6. Risk of bias and quality of methods assessment

The risk of bias was assessed using the Cochrane Collaboration's Risk of Bias tool 2 (Rob 2),²⁵ which evaluates random sequence generation, random allocation concealment, blinding of outcome assessment, incomplete outcome data, and selective outcome reporting. Disagreements were resolved through discussion whenever possible. If consensus could not be reached, a third reviewer acted as an arbitrator. Additionally, the physiotherapy evidence database (PEDro) scale was used to assess methodological quality.²⁶ Scores range from 0 to 10, with studies classified as high quality (≥6), moderate quality (4–5), or low quality (≤3).

2.7. Statistical analysis

We first applied a traditional two-level meta-analysis based on a generic inverse-variance pooling method to pool Hedges' g and were conducted using the meta and metafor packages in the statistical software R (V.4.2.0).²⁷ For the two-level meta-analysis, we utilized the DerSimonian-Laird approach,²⁸ which is a random-effects model accounting for potential heterogeneity across studies. This model assumes that effect sizes are derived from a distribution of true effects rather than from a single homogeneous population. Given the variation in training protocol and populations, the random-effects model incorporates heterogeneity by assuming that the underlying effects follow a normal distribution, leading to a more accurate and appropriate estimation of the overall effect size.¹⁵

In cases where studies included nested or multiple effect sizes (e.g., for outcomes such as jump performance), these effect sizes were often correlated. Including all effect sizes simultaneously could violate the assumption of independence in traditional meta-analyses,²⁹ while considering only one effect size could be too conservative and fail to capture the true effect.³⁰ To address dependency structures, we applied a three-level meta-analysis following the methods of Pustejovsky & Tipton,³¹ we first constructed an approximate variance-covariance matrixV using the vcalc function to account for overlapping samples. We then fitted a three-level model with rma. mv and applied cluster-robust inference via the robust function with the clubSandwich package for improved small-sample adjustments. Sensitivity analyses using cluster wild bootstrapping (wildmeta package) were conducted when degrees of freedom were critically low. This approach allows for the decomposition of variance into sampling variance (level 1), within-study variance (level 2), and between-study variance (level 3), accounting for correlated and hierarchical effects.³² By preserving valuable information from multiple effects within each study, the three-level meta-analysis enhances statistical power and provides a more accurate representation of effect sizes.³³ For the three-level model, parameters were estimated using restricted maximum likelihood (REML) estimation with the rma. mv function in the R metafor package, and the results were cross-verified using the maximum likelihood method to ensure stability.

We calculated 95 % confidence intervals (CIs) using the Knapp-Hartung adjustment (test = ‘knha’) with t-distributions for individual coefficients and F-distributions for omnibus tests. For multilevel models fitted with rma. mv, degrees of freedom were approximated viadfs = ‘contain’. Additionally, we computed the prediction interval (PI) for metrics with >5 included studies based on the t-distribution, which measures the treatment effect considering heterogeneity and provides useful additional information compared to the CI and used to estimate the range of the overall parameter and to account for the uncertainty of future observations,³⁴ especially considering the use of a random-effects model.^35,36 The between-study variability (i.e., heterogeneity) of the intervention effects within each intervention comparison was assessed by I^{2 37} and the magnitude of the between-study variance (τ2) estimated using the generalized DerSimonian and Laird³⁸ estimator and the Q-profile approach. Therefore, the main analysis reports I² with the following interpretations: 0–25 %, might not be important; 25–50 %, may represent moderate heterogeneity; 50–75 %, may represent substantial heterogeneity; and 75–100 %, considerable heterogeneity.¹⁵ Additionally, the statistical power of the primary pooled effect was calculated, and the possibility of false negatives due to insufficient statistical power was considered. Statistical power calculations were performed using the metameta package.³⁹

Potential sources of heterogeneity and moderators were explored by subgroup analyses. The following variables were selected for the subgroup analyses: (1) testing metrics (unilateral performance or bilateral performance); (2) training frequency; (3) training protocol.

2.8. Risk of publication bias and sensitivity analysis

A contour-enhanced funnel plot,⁴⁰ along with Egger's asymmetry test^41,40 was employed to assess publication bias. These tests were only conducted when the number of included studies (k) was ≥10,⁴² and a p-value > 0.05 was interpreted as indicating no evidence of publication bias. Funnel plots and Egger's regression tests assess the symmetry of the overall effect size, either subjectively or quantitatively, to evaluate the risk of publication bias in the included studies.

Sensitivity analyses were conducted using a leave-one-out approach, sequentially removing each study to assess its influence on the overall pooled effect. Cluster-robust variance estimation methods⁴³ with small-sample adjustments⁴⁴ were applied to adjust the within-study standard errors for correlations between effect sizes. If the results changed significantly, we applied these methods; otherwise, we retained the original model.

2.9. Certainty of the evidence

The risk of bias was considered in the interpretation of the results by applying the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) methodology, which rates the certainty of evidence as “high”, “moderate”, “low” or “very low”.⁴⁵ GRADE assessments were initially completed by one reviewer and verified by a second reviewer for verification.

3. Results

3.1. Search results

A flow diagram of the study selection process is presented in Fig. 1. Overall, 10,829 studies were identified in the initial database search. Following the removal of duplicates (n = 4581), 6248 titles and abstracts were screened against the inclusion criteria, and 6180 studies were irrelevant. A full-text review of the remaining 68 studies excluded a further 60 studies for the following reasons: not a randomized controlled trial (n = 11), not involving bilateral or unilateral intervention (n = 21), not related to a competitive sports population (n = 18), and outcome indicators not met (n = 10). Subsequent screening resulted in 8 papers.46, 47, 48, 49, 50, 51, 52, 53 Following this, a rolling snowball secondary search was conducted, which identified and incorporated one additional study that met the inclusion criteria and was ultimately included in the meta-analysis⁵⁴ (Fig. 1) (see Fig. 2).

Fig. 1.

PRISMA flow diagram of the search strategy.

Fig. 2.

Pooled effect sizes for the outcomes.

Note: K, the total number of effects included in the pooled effect size; Hedges' g, the effect size indicators used in the pooled; 95 %PI, 95 % prediction interval; P-value, statistically significant P values for pooled results; I²-2, quantitative indicators of heterogeneity for two-level; I²-3, quantitative indicators of heterogeneity for three-level.

3.2. Study characteristics

A total of 225 participants were included in these nine studies. Among the eight studies that reported gender, there were 12 female participants, 195 male participants, and 18 participants whose gender was not reported. A detailed description of the study participants is presented in Table 1, with a mean age of 15.28 ± 3.10 years across the studies. All participants in the included studies were athletes (from sports such as soccer, rugby, basketball, and track and field). In nine studies, the training programs were supervised by members of the research team or physical therapists. The mean training duration was 7.56 ± 4.07 weeks (ranging from 5 to 18 weeks). The training frequency was two times per week in seven studies and three times per week in two studies.

Table 1.

Characteristics of the included studies.

Study	Study design	Sample	Age (yr)	Gender	Sports	Training	Training duration (weeks)	Frequency	Training volume between the groups
Ramírez-Campillo et al., 2015	RCT	54	11.4 ± 2.2	M	soccer	Plyometric Training	6	2	UT: BT
Speirs et al., 2016	RCT	18	18.1 ± 0.5	n/a	rugby	Strength Training	5	2	two times UT: BT
Gonzalo-Skok et al., 2017	RCT	22	16.9 ± 2.1	M	basketball	Strength Training	6	2	two times UT: BT
Ramírez-Campillo et al., 2018	RCT	18	17.45 ± 0.8	M	soccer	Strength and Plyometric Training	8	2	two times UT: BT
Appleby et al., 2020	RCT	33	15.6 ± 1.0	M	rugby	Strength Training	18	3	two times UT: BT
Stern et al., 2020	RCT	23	17.6 ± 1.2	M	soccer	Strength Training	6	2	two times UT: BT
Cillík et al., 2023	RCT	16	12.6 ± 1.5	M/F	athletic	Strength and Plyometric Training	8	3	UT: BT
Zhao et al., 2023	RCT	26	15.3 ± 0.4	M	rugby	Strength Training	5	2	two times UT: BT
Fisher and Wallin, 2014	RCT	15	19.96 ± 1.58	M	rugby	Strength and Plyometric Training	6	2	two times UT: BT

Note: RCT, randomized controlled trial; M: male; F: female; yr: years; Frequency, training frequency (sessions/week); UT: unilateral training; BT: bilateral training.

In this systematic review, we included studies that assessed the training volume between UT and bilateral training BT. The UT: BT indicates that the training volume applied to both groups in UT and BT is equivalent (e.g., if BT performs 3 sets of 6 repetitions at 75 % 1RM, then UT performs 3 sets of 3 repetitions at 75 % 1RM, making the total training volume for UT and BT equal). In contrast, two times UT: BT suggests that the training volume of UT is twice that of BT (e.g., if both UT and BT perform 3 sets of 6 repetitions at 75 % 1RM, the UT volume doubles because each side completes 6 repetitions). The training protocols in these studies included unilateral and bilateral strength training, strength training combined with plyometric training, and pure plyometric training. Training frequencies were either two or three times per week. Using these data, we will conduct subgroup analyses focusing on the testing metrics (unilateral vs. bilateral performance), training frequency, and training protocols.

3.3. Methodological quality of included studies

The risk of bias for each study is presented in Electronic Supplementary Material Appendix S2 Risk of Bias. Most studies did not report detailed information regarding their randomization methods and allocation concealment, leading to a rating of “some concerns” for the randomization process. Additionally, two studies reported dropout rates among participants exceeding 15 %. In summary, most studies demonstrated a “some concerns” risk of bias.

The PEDro scores obtained ranged from moderate to high quality, with a mean score of 6.90 for this systematic review and meta-analysis. Therefore, the overall quality was rated as high. A detailed summary of the methodological quality assessment, including individual PEDro scores for each study, is provided in Electronic Supplementary Material Appendix S5 Methodological Quality Assessment.

3.4. Certainty of the evidence

The risk of bias was considered in the interpretation of the results by applying the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) methodology, which rates the certainty of evidence as “high”, “moderate”, “low” or “very low”.⁴⁵ GRADE assessments were completed by one reviewer and reviewed by a second.

3.5. Primary analysis

We conducted a pooled the included UT and BT studies to analyze their effects on athletic performance. Separate analyses were conducted for studies with group-specific loading characteristics of UT: BT and two times UT: BT to examine their impact on performance. The main effect results showed no significant difference in improving athletic performance when the loading settings of UT and BT were identical (K = 13, N = 316, g = −0.20, 95 % CI [−1.56, 1.15], p = 0.75, I²-2 = 77.8 %; I²-3 = 0.0 %, PI [−4.54, 4.14], low GRADE). Additionally, there was no significant difference when two times UT and BT (K = 34, N = 653, g = −0.04, 95 % CI [−0.14, 0.06], p = 0.33, I²-2 = 93.8 %; I²-3 = 0.0 %, PI [−1.96, 1.87], moderate GRADE) (Electronic Supplementary Material Appendix S6 Main analyses).

A visual plot of statistical power for the pooled results for all outcomes in Electronic Supplementary Material Appendix S1.

3.6. Moderator analysis

When the training volume was the same between UT and BT (UT: BT), athletic performance was not significantly moderated by testing metrics, training frequency, or training protocol. Regarding testing metrics, both unilateral performance (g = −0.84) and bilateral performance (g = −0.90) showed substantial but non-significant improvements. Regarding training frequency, neither two times per week (g = −0.16) nor three times per week (g = −0.19) had a significant improvement on athletic performance. Regarding training protocols, plyometric UT and BT (g = −0.16) and strength combined with plyometric UT and BT (g = −0.19) did not significant improvement in athletic performance.

When the UT training volume was twice that of BT (two times UT: BT), athletic performance was significantly moderated by testing metrics but not by training frequency or protocol. Regarding testing metrics, unilateral performance improved significantly (g = −0.39), while bilateral performance showed a significant improvement (g = 0.64) on athletic performance. Regarding training frequency, neither two times per week (g = −0.14) nor three times per week (g = 0.13) had a significant improvement on athletic performance. Regarding training protocols, strength UT and BT (g = −0.01) and strength combined with plyometric UT and BT (g = −0.21) did not show significant improvement on athletic performance.

3.7. Risk of bias and sensitivity analysis

The risk of publication bias was investigated using a funnel plot combined with Egger's test for the effects of included studies on UT: BT and two times UT: BT studies. T No publication bias was observed in UT: BT (p = 0.12), while publication bias was observed in two times UT: BT (p = 0.03) (Electronic Supplementary Material Appendix S4 Funnel Plot).

We conducted sensitivity analyses using a leave-one-out method for all primary pooled effects (Electronic Supplementary Material Appendix S5 Leave-one-out). The results indicated that the exclusion of any single study did not significantly alter the pooled outcome. This suggests that our findings are robust and reliable.

4. Discussion

This systematic review and meta-analysis aimed to explore the effects of training volume settings between UT and BT on athletic performance improvements. The primary findings revealed that when UT and BT were implemented with equal training volumes, neither training showed substantial advantages in enhancing athletic performance, with outcomes remaining consistent across testing metrics, training frequencies, and training protocols. Similarly, even when UT training volume was doubled compared to BT, no significant overall superiority of either training was observed. However, a critical moderating effect observed for testing metrics: two times UT volume for BT selectively improved unilateral performance outcomes (e.g., single-leg jump, unilateral strength), whereas BT maintained its efficacy for bilateral performance (e.g., bilateral squats). Training frequency and protocol did not significantly moderate athletic performance.

When UT and BT were applied with equal or doubled training volumes, neither training showed clear superiority in enhancing athletic performance. This lack of differentiation may stem from our aggregated analysis of unilateral and bilateral performance data, which could have obscured the principle of specificity in the main effects. For example, Liao et al.² demonstrated that UT preferentially improves unilateral jump performance, while BT enhances bilateral strength performance. However, this limitation was effectively resolved through subgroup analyses, which enabled a more nuanced exploration of the distinct effects of UT and BT by isolating performance metrics aligned with their inherent specificity.

Regarding testing metrics showed a consistent trend across both UT: BT and two times UT: BT: UT induced greater improvements in unilateral performance (e.g., single-leg jump, sprinting), while BT more prominently improved bilateral performance (e.g., bilateral squat). However, these differences reached statistical significance only when UT volume was doubled for BT. This highlights the role of training volume difference between UT and BT on athletic performance. The significant improvement in unilateral movements may be attributed to the improved neuromuscular adaptations from the increased unilateral load,⁴⁶ including improved knee joint stability and optimized hamstring-to-quadriceps activation ratios.¹⁷ Additionally, task-specific matching plays a crucial role, as unilateral performance (e.g., change of direction, sprinting) relies on rapid unilateral force production, which UT directly enhances through single-joint dominant training.⁵⁵ Conversely, BT, by simultaneously activating bilateral muscle groups, more effectively improves performance in tasks requiring coordinated bilateral force output.⁵⁶

Regarding training frequency, no significant differences were observed between two times per week and three times peer week. This suggests that increasing training frequency may not provide additional benefits, and that optimal load distribution combined with adequate recovery time could play a more critical role in performance enhancement.⁵⁷ Finally, regarding the training protocols, we found no significant differences between plyometric training (g = −0.16) and strength and plyometric training (g = −0.19) in the UT: BT. However, in the two times UT: BT, strength and plyometric training (g = −0.21) showed a larger effect size compared to strength training (g = −0.01). The superior efficacy of strength and plyometric training in improving muscular strength may be attributed to the combination of these modalities, which is likely to result in greater power output.⁵⁸ and improvements in joint control strategies and/or the rate of torque development in the knee extensors.⁵⁹

4.1. Practical applications

It is recommended to implement strength training combined with plyometric training for two or three times per week, using two times the volume of unilateral training compared to bilateral training (two times UT: BT) to improve unilateral or bilateral athletic performance.

4.2. Potential limitations

Although this meta-analysis provides new insights into the enhancement of athletic performance through UT and BT, several limitations need to be noted. First, our inclusion criteria encompassed nine studies published in English. The small number of studies limited the statistical power and precision of our pooled effect sizes and subgroup analyses, but this potential risk of bias is unlikely to influence our pooled effect size. Secondly, differences in participant characteristics and testing methods have resulted in heterogeneity between the UT and BT protocols. Although we addressed this potential heterogeneity through subgroup and sensitivity analyses, caution is still warranted in interpreting the study results.

5. Conclusion

The training volume settings between unilateral and bilateral training may not have a significant impact on athletic performance. The testing metrics might be the significant moderating factors, whereas training frequency and training protocol are likely not significant moderators.

Ethical statement

This review article does not involve human or animal experiments, and it is based exclusively on published literature. Therefore, a Statement of Ethics is not applicable.

Funding source

No sources of funding were used to assist in this article.

Conflict of interest statement

No conflicts and interests are relevant to the content of this review.

Appendix A. Supplementary data

The following is the Supplementary data to this article: