Eligibility criteria

We defined the eligibility criteria in line with the Population – Concept – Context (PCC) framework developed by the Joanna Briggs Institute and recommended for scoping reviews [22].

Population

Inclusion: Soccer players of any age, sex, or competitive level. Studies must involve healthy participants who are actively engaged in training or competition.

Exclusion: Non-athlete populations (e.g., physical education students without organised training, recreational adults not involved in structured sport) and athletes from sports other than soccer. Clinical or injured players are also excluded.

Concept

Inclusion: Studies explicitly examining and comparing training regimens within SSGs, including: Continuous formats (uninterrupted play of fixed duration); Intermittent formats (play interspersed with predefined recovery periods); Intermittent formats with different work-to-rest ratios (e.g., 1:1, 1:2). SSGs implemented as structured training interventions, whether acutely (single session) or longitudinally (multiple sessions/seasonal). Game format (e.g., 3v3–7v7), the presence of a goalkeeper, and pitch dimensions were not used as inclusion or exclusion criteria. Outcomes of interest: Physical (e.g., external load measured by means of distances covered or accelerations or compounds of variables); Psychophysiological (e.g., HR, RPE, hormonal or metabolic responses); Technical (e.g., passing accuracy, number of technical actions); Tactical (e.g., team organisation, positional and spatial-temporal variables).

Exclusion: Studies focusing only on other SSG task constraints (playing area, number of players, rules modifications) without manipulating the training regimen. SSGs used solely as an assessment or testing tool, not as a training stimulus. Studies where SSGs are combined with other interventions (e.g., resistance training, sprint training), unless the isolated effect of the regimen structure can be discerned. Studies reporting exclusively psychological (e.g., enjoyment, motivation) or sociological outcomes without physical, psychophysiological, technical, or tactical data.

Context

Inclusion: Training-based studies conducted in competitive sport contexts (e.g., grassroots, club, academy, professional team environments). No restriction on geographic location or competitive season.

Exclusion: Non-sport contexts (e.g., general education or recreational programmes for older adults).

Types of sources

Inclusion: Original empirical research of any design (randomised controlled trials, quasi-experimental studies, crossover, repeated-measures, cross-sectional, or longitudinal/observational) was eligible. In principle, both cross-sectional studies (examining acute responses within single sessions) and longitudinal studies (evaluating training adaptations across multiple sessions or weeks) were considered, as they provide complementary information. However, all included studies in this review were designed to examine the acute, sessional effects of different SSG regimens. Although some protocols involved repeated bouts, sets, or testing sessions, no study investigated longitudinal training adaptations over weeks or months. Accordingly, all included evidence was classified as cross-sectional. Full-text articles published in peer-reviewed journals with adequate methodological detail were considered, with no restrictions on publication year or language.

Exclusion: Reviews, meta-analyses, editorials, commentaries, opinion papers, book chapters, theses, conference abstracts/articles, and dissertations. Abstract-only publications without sufficient methodological detail.

Information sources and search strategy

Searches were conducted in PubMed, Scopus, and Web of Science (Core Collection). These sources were selected as they represent some of the most relevant databases for sports science, exercise physiology, and sports medicine literature. The final searches were conducted on 28 August 2025. To ensure the evidence base was as complete as possible, additional searches of grey literature were carried out on Google Scholar. Reference lists of all included studies and related reviews were manually screened to identify further eligible articles, and a snowballing process was applied to capture additional studies through forward and backward citation tracking. Moreover, domain experts in SSGs and sport science were consulted to verify the comprehensiveness of the search and to identify studies that may not have been retrieved through database searching.

The search strategy was developed by combining free-text terms with Boolean operators to optimise retrieval. The final search string was:

[Title/Abstract] ("small-sided game" OR "small-sided games" OR "SSG" OR "SSGs" OR "conditioned game" OR "conditioned games" OR "reduced game" OR "reduced games" OR "modified game" OR "modified games" OR "sided-game" OR "sided-games" OR "medium-sided game" OR "medium-sided games" OR "large-sided game" OR "large-sided games" OR "constrained game" OR "constrained games")

AND

[Title/Abstract] ("training regimen" OR "continuous" OR "intermittent" OR "work-to-rest ratio" OR "interval" OR "training format" OR "fractionated" OR "recovery" OR "rest")

The search syntax was adapted to the specific requirements of each database. In addition to the database searches, grey literature was screened through Google Scholar (first 200 hits per query, retrieved on 28 August 2025). Forward and backward citation tracking of all included records was also conducted on 29 August 2025. All additional records identified through these procedures were screened using the same PCC criteria.

Selection of sources of evidence

All records identified through the database and supplementary searches were exported into the End-Note online software, and duplicates were removed prior to screening. The selection of sources of evidence was performed in two stages. First, titles and abstracts were independently screened by two authors (FMC and DM) against the eligibility criteria. Second, the full texts of potentially relevant articles were retrieved and assessed in detail for final inclusion. Any disagreements between authors at either stage were resolved through discussion, and when consensus could not be reached, a third author acted as arbiter (RT).

The entire selection process was documented in accordance with the PRISMA-ScR flow diagram, reporting the number of records identified, screened, excluded (with reasons), and included in the final synthesis. Reasons for exclusion at the full-text stage were systematically recorded.

Data charting process

Data from the included sources of evidence were charted using a standardised extraction form developed a priori in line with the objectives of this review. The form was designed iteratively: an initial version was piloted independently by two authors (FMC and DM) on a random sample of five studies. Based on this pilot phase, adjustments were made to refine the definitions of data items and to harmonise the charting procedure.

Following piloting, two authors (FMC and DM) independently charted data from all included studies. Extracted information covered bibliographic details (author, year), study characteristics (design, sample size, sex and competitive level of participants), details of the SSG training regimen (continuous, intermittent, or intermittent with varying work-to-rest ratios), session parameters (duration, frequency, total intervention period), and SSG design (e.g., format of play [e.g., 3v3, 4v4], pitch dimensions, task objectives, presence of goalkeeper (yes/no) and other task conditions). Operational definitions for training regimens were: (i) Continuous SSGs: Games played for a predetermined duration without planned intra-set rest intervals; (ii) Intermittent SSGs: Games structured in bouts separated by planned rest intervals (e.g., 4 × 4 min with 2 min rest); (iii) Density: The within-set work–rest structure, defined as the work-to-rest ratio (work duration ÷ rest duration), reported as x:y or as a single quotient.

Any discrepancies in data extraction between authors were resolved through discussion. When important information was missing or unclear, attempts were made to contact study authors to obtain additional details through e-mail and the ResearchGate platform.

Data items

Outcome measures were grouped into four categories aligned with the objectives of this review. Physical outcomes included external load indicators such as total distance (TD); distance covered in standardised speed zones [e.g., walking, jogging, moderate-intensity running (MIR), high-intensity running (HIR), very high-intensity running (VHIR), sprinting]; distance above test-based thresholds [e.g., > 80% of peak speed from Carminatti’s intermittent running test (PST-CAR)] or fixed cut-offs (e.g., 13, 17, 18, 21 km • h^–1); maximal velocity (Vmax); accelerations (ACC) and decelerations (DEC); per-minute expressions (m • min^–1); combined load metrics such as PlayerLoad (PL, accelerometer-derived composite workload); GPS-derived time–motion data (TMD); and repeated sprint or metabolic power indices.

Psychophysiological outcomes comprised heart rate (HR) measures [mean, peak, percentage of maximum heart rate (%HRmax), and time spent in HR zones]; blood lactate concentration (La^–); ratings of perceived exertion (RPE) and sessional ratings of perceived exertion (sRPE, Borg CR-10 scale); enjoyment or mood scales; internal training load indices such as training impulse (TRIMP); and hormonal or metabolic markers, including total testosterone (TT), free testosterone (FT), cortisol (C), testosterone-to-cortisol ratio (T/C), and free testosterone-to-cortisol ratio (FT/C).

Technical outcomes included the number of passes, successful versus unsuccessful passes, pass accuracy (%), number of shots (on and off target), goals scored, dribbles, tackles, turnovers, goal-scoring opportunities, ball contacts, and ball involvements.

Tactical outcomes were derived from notational or positional analyses and included team surface area, pitch coverage, stretch index, inter-player synchronisation, team dispersion, player interactions, and spatial– temporal patterns.

In this review, technical outcomes were defined as discrete, individual-level actions (e.g., passes, shots, tackles) typically captured through notational analysis, while tactical outcomes were defined as collective behaviours derived from positional data (e.g., team dispersion, synchronisation, stretch index). We recognise that several validated observational instruments, such as the Game Performance Evaluation Tool [23] and the Football Assessment System (FUT-SAT) [24, 25], provide tactical indicators without positional tracking; however, no eligible study in our sample employed these tools in the context of comparing training regimens.

Synthesis of results

Data were organised into evidence tables that mapped the extent, range, and nature of the literature examining different training regimens in SSGs. Studies were grouped according to the format of play (continuous, intermittent, or intermittent with varying work-to-rest ratios) to allow comparisons across training regimens. Within each category, findings were further classified according to outcome domains: physical, psychophysio-logical, technical, and tactical. Two authors (FMC and DM) independently coded each included study into outcome domains (physical, psychophysiological, technical, and tactical). Disagreements were resolved by consensus. For studies reporting outcomes across multiple domains, all relevant domains were coded.

Patterns of evidence were summarised narratively, consistent with the objectives of a scoping review, with emphasis on convergent results, conflicting findings, and areas of insufficient evidence. Particular attention was paid to how variations in training regimens influenced responses across outcome domains, as well as to methodological gaps. The synthesis also incorporated visual mapping of evidence through figures and tables to illustrate the distribution of studies across regimens, outcomes, and populations.

Selection of sources of evidence

A total of 2,018 records were identified through database searches (PubMed, n = 406; Scopus, n = 843; Web of Science, n = 769). After the removal of 948 duplicates, 1,070 records remained for screening. Of these, 1,034 were excluded based on title and abstract. Thirty-six (36) full-text reports were retrieved and assessed for eligibility, with none being unobtainable. Following the assessment, six (6) studies [26, 31] were excluded because they did not include comparisons between training regimens. Ultimately, thirty (30) studies met the inclusion criteria and were incorporated into the review (Figure 1).

Figure 1

PRISMA flowchart

Characteristics of sources of evidence

Across the 30 studies synthesised in Table 1 (see end of paper), most comparisons focused on continuous versus intermittent SSG formats [10, 14, 32–35]. Comparisons between intermittent formats manipulating the bout duration were also frequently reported [13, 26, 36]. Additionally, studies examined the effects of different recovery times, with workout density emerging as a common variable of interest [37, 38]. Psycho-physiological outcomes, especially HR, RPE, and lac-tate concentration, were the most consistently reported measures in the literature [10, 39], whereas technical actions were scarcely quantified [32, 34].

The evidence gap map (Figure 2) highlights this imbalance, showing dense coverage in the physical and psychophysiological domains but considerably fewer studies addressing technical or tactical outcomes. In reality, tactical measures of collective behaviour remain almost absent. Moreover, although work has begun exploring metabolic stress markers such as acid– base balance in high-intensity 1v1 SSGs [40, 41] and mood-related outcomes [42], these remain underrep-resented in the broader dataset.

Figure 2

Evidence gap map of training regimen comparisons by outcome domain. Numbers in cells indicate the number of studies contributing to each cell.

Analysis of study designs by format of play showed that most evidence was accumulated in 4v4, 3v3, and 5v5 SSGs, particularly when comparing continuous and intermittent regimens (Figure 3). For example, 9 studies contrasted continuous and intermittent structures in 4v4 formats, while 7 and 4 studies did so in 3v3 and 5v5, respectively. In contrast, evidence for larger formats (6v6, 7v7) or very small-sided designs (1v1, 2v2) remains unstudied, with only isolated comparisons reported. Intermittent–intermittent contrasts (longer vs shorter bouts or longer vs shorter recovery) have been studied in 3v3 to 5v5 formats but are virtually absent in 6v6 and 7v7.

Figure 3

Evidence gap map considering the type of training regimen comparisons and the formats employed. Numbers in cells indicate the number of studies contributing to each cell.

Results of individual sources of evidence

Comparisons between continuous and intermittent regimens (Table 2, see end of paper) consistently indicate that intermittent formats tend to elicit greater distances covered at high speeds and more accelerations, while continuous play sustains higher cumulative workloads [10, 32, 43]. The work-to-rest ratio was also a critical determinant: longer recoveries allowed greater high-intensity outputs, whereas shorter rests constrained physical performance [38, 44, 45].

Across formats, continuous regimens generally produced higher mean HR and RPE [43, 46, 47]. Conversely, intermittent regimens often elicited similar or higher HR peaks but lower sessional RPE (sRPE) [10, 35, 48]. Manipulating the bout and rest durations showed consistent patterns: shorter work-to-rest ratios increased lactate and RPE [38, 39], whereas longer recoveries attenuated internal load and improved metabolic recovery [37, 44, 45]. Some recent studies also expanded into hormonal [49] and mood-related markers [42]. Table 3 (see end of paper) presents the main findings, showing how different bout and recovery structures distinctly shape cardiovascular, metabolic, and perceptual demands in SSG training.

Continuous regimens often favoured passing accuracy and ball possession (Table 4, see end of paper), while intermittent regimens tended to increase the number of duels, tackles, and goal-scoring opportunities [43, 42, 46]. Very short recoveries reduced passing success, whereas longer rests allowed better execution of ball actions [38, 45].

Continuous vs intermittent regimens

Across the studies that directly contrasted continuous with intermittent SSGs, intermittent formats often showed higher external intensity (e.g., more high-intensity running/sprints or greater distance per minute), whereas continuous formats more often sustained steadier workloads, although the effects were context-dependent. Intermittent formats were more intense than continuous for high-intensity running and sprint actions in young players while performing 2v2–6v6 [43], producing slightly greater total distance than continuous in 5v5 possession [10], and maintained higher external outputs when very short work bouts were used (e.g., 6 × 3 vs 3 × 6 in 5v5; [11, 13]). This pattern is consistent with the underlying physiology, as rest intervals in intermittent play likely permit partial phosphocreatine resynthesis and metabolite clearance, enabling repeated high-intensity efforts, whereas continuous play sustains a steadier cardiovascular and perceptual load across the bout [50–52].

Intermittent formats elicited higher-intensity running and sprint actions compared to continuous formats, likely due to the partial recovery periods allowing repeated maximal efforts, while in continuous formats, players may self-regulate their intensity to sustain efforts over the longer duration, resulting in lower peak actions. Pedagogically, shorter bouts can also increase the density of action – perception cycles, promoting decision-making opportunities and engagement, while longer rest intervals may preserve technical execution (e.g., passing accuracy) by reducing fatigue-related errors [53]. In the applied contexts, ‘intensive’ SSGs formats (shorter bouts/blocks) sometimes delivered less running volume than ‘extensive’ (longer bouts) but altered the load profile [54]. In 6v6, however, continuous, 2 × 4 min, and 4 × 2 min showed no clear differences in external load [34].

In the psychophysiological analysis, continuous regimens frequently elicited higher mean HR and RPE than intermittent [34, 43, 46, 55], while intermittent formats sometimes showed higher HR peaks but lower sRPE or more fluctuating intensity [10, 35, 48]. Longer continuous bouts were often perceived as harder than fractionated equivalents even when the external loads were similar [34], and ‘intensive’ sessions were associated with higher sRPE and ITL than ‘extensive’ sessions [54]. Over weeks, intermittent SSG improved anaerobic power more than continuous SSG, while continuous involved a higher perceived load [35, 56]. Intermittent formats increased shots/goals vs continuous in 6v6 [34], yet continuous play was associated with better passing accuracy and lower ball loss in 3v3–4v4 formats [42, 46].

Shorter vs longer intermittent workout periods

When the work duration within intermittent SSGs was manipulated while keeping the recovery relatively constant, shorter bouts tended to maintain higher external intensity. In 5v5 training, six 3-min bouts elicited more total distance, running distance, and accelerations/decelerations compared to three 6-min bouts [11, 13]. Similarly, in youth players, shorter repetitions (6 × 2 min) increased distance per minute and moderate-intensity running compared with longer sets (2 × 6 min), while long-bout conditions maintained higher total volumes [55]. In the 1v1 format, 30-s work intervals produced greater relative distances and higher maximal velocities than 45-s intervals, though overall work output was reduced [40, 41]. Still, not all comparisons revealed differences. In the 6v6 format, intermittent formats of varying bout lengths (8 min, 2 × 4 min, 4 × 2 min) showed no clear differences in total distance, high-speed running, or accelerations [34]. The number of players likely modulates both the magnitude and orientation of the physical response in SSGs. As player numbers increase, tactical constraints and reduced individual involvement may attenuate the physical demands, largely independent of the bout duration. This suggests that the format itself, rather than interval length alone, can play a key role in shaping training responses.

When including goalkeepers in 5v5, three 6-min bouts resulted in higher RPE values than six 3-min bouts, even though the mean HR did not differ significantly [11, 13]. Similar results were observed in 4v4, where three 6-min formats produced higher mean and peak HR compared to six 3-min [36]. Lactate responses also tended to rise more in long-bout formats [55]. Regarding the technical aspect, shorter bouts were associated with a greater frequency of individual actions and more stable execution [32]. In contrast, longer bouts allow the technical component to work under fatigue, still with higher rates of passing mistakes or ball losses, suggesting the possibility of using this constraint to represent the variability of the last periods of an official soccer match [42, 46].

From a skill-acquisition standpoint, higher action frequency provides more practice opportunities; in general motor-learning research, distributing practice (i.e., inserting rest to maintain quality across reps) improves long-term retention compared with massed practice [57]. Likewise, varying actions across bouts (contextual interference) tends to enhance retention/ transfer, supporting the use of short, intensive bouts to accumulate high-quality, variable technical repetitions [58, 59]. Consistent with these principles, SSG training programs have been shown to improve technical execution in youth players over time, indicating that well-designed formats can translate acute repetition into longer-term skill gains [60, 61]. Conversely, because fatigue degrades technical skill (e.g., passing and shooting accuracy), deliberately practicing within longer, more fatiguing bouts can target robustness of execution under match-relevant constraints (specificity) [62, 63]. This aligns with nonlinear-pedagogy guidance to manipulate task constraints (e.g., bout and rest structure) to develop decision-making and performance in representative environments [64, 65].

Shorter vs longer rest periods

When the recovery duration between intermittent SSG bouts was manipulated, longer rests generally allowed players to sustain or increase high-intensity outputs, whereas shorter rests tended to constrain external load. In the 3v3 format, 3-min rests increased total distance and high-intensity running, while 4-min rests promoted more moderate-intensity activity compared with shorter pauses [38]. Similarly, in 4v4 SSGs, extending the rest to 4 min produced more accelerations and decelerations compared to 1-min recovery, although total distance, high-speed running, and maximal velocity were unaffected [37]. In contrast, and in 6v6 format, varying the recovery from 30 s to 120 s between 2-min bouts did not significantly alter the total distances or speed zones, suggesting that players may adapt pacing strategies when rest windows are relatively short [44, 45], or that this recovery duration within this range may be insufficient to meaningfully influence external load responses.

Shorter rests consistently increased cardiovascular and metabolic stress, whereas longer rests attenuated these responses. In 3v3, 1-min rest produced higher %HRmax, greater blood lactate concentration, and higher RPE compared with 2–4 min pauses [38]. In 4v4, 2-min recovery was associated with higher RPE than 1-min, while 4-min did not further increase the perceived exertion [37]. In 6v6, HR during exercise was similar between 30-s and 120-s rest, but recovery HR was significantly lower with the longer pause [44, 45]. Regarding technical outcomes, the evidence is limited but suggests that very short rests may impair execution (similar to longer bouts). In 3v3 formats, 1-min recovery was linked to reduced successful passes, while 4-min rests increased passing success and 3-min rests favoured tackles and ball touches [38]. McLean et al. [45] also noted that 120-s rests supported more successful tackles compared with 30-s intervals.

Limitations and future research

Despite the important findings of this scoping review, there is a need to acknowledge some limitations. First, the heterogeneity in study designs, reporting formats, and outcome measures limited the ability to synthesise the results beyond descriptive mapping. Second, common methodological limitations were evident across the included studies, such as small and often homogeneous samples (mostly male, youth, or sub-elite players), short intervention periods, and a lack of longitudinal approaches. Many studies also relied on isolated sessions rather than longitudinal designs, and only a minority reported the reliability of the outcome measures. To advance this research topic, future studies should prioritise larger and more diverse samples, including female or elite players. Longer intervention studies are needed to move beyond acute responses and better capture adaptation.

Another important limitation concerns the restricted assessment of technical and tactical outcomes, which were seldom quantified and, when reported, often relied on subjective or low-resolution methods. Several factors may explain this. First, measurement is less straightforward than for physical or psychophysiological variables, as technical actions (e.g., passing accuracy, dribbles) typically require manual coding, which is both time-intensive and prone to observer bias. Second, there is no consensus on measurement approaches: some studies report absolute frequencies of actions, others focus on success rates, while still others apply composite indices, making direct comparisons difficult. Finally, the absence of widely adopted, validated tools for assessing technical and tactical skills has led to considerable variation in how ‘technical performance’ and ‘tactical performance’ are defined and measured, resulting in inconsistent findings. These limitations underscore the need for more standardised approaches to technical and tactical assessment in SSG research. Furthermore, contextual factors such as coaching strategies, task constraints, or motivational variables were largely unreported, despite their potential influence on responses to SSGs. Clearer descriptions of training regimens (e.g., playing area per player, rest type, rules, task goals) and replicability conditions would help establish findings that can better develop practice.

Finally, the inclusion of psychological and social variables may be relevant for future studies. For example, enjoyment and motivation are important moderators of training engagement and long-term adherence [66], and they may vary depending on the bout structure or work-to-rest ratio. Similarly, communication and cooperation between players influence tactical coordination and technical execution [67], yet are rarely quantified alongside physical or technical indicators. Linking these variables with dominant outcomes (e.g., relating enjoyment to external load, or communication frequency to passing accuracy) would provide a more ecological overview of how different SSG regimens shape the holistic training experience.

Beyond the limitations of the included studies, several constraints of this review should be acknowledged. Our scope was restricted to the acute, session-level effects of SSGs, and none of the included studies tracked longitudinal adaptations. As such, conclusions about training-induced changes over time cannot be drawn from this synthesis and remain an important avenue for future research. Moreover, this review followed a scoping methodology, which is designed to map the nature and extent of research rather than to appraise the risk of bias or conduct quantitative synthesis. While this approach is appropriate in a field characterised by heterogeneity in study designs, outcomes, and measurement tools, it limits the ability to draw conclusions about effect magnitude or causal certainty. A systematic review incorporating risk-of-bias assessment and, where feasible, meta-analysis would be warranted once methodological heterogeneity is reduced, training regimens can be more clearly isolated, technical and tactical outcomes are more consistently stand-ardised, and longitudinal interventions become more available.

Practical applications

The findings of this scoping review can offer tentative considerations for coaches and practitioners, provided the heterogeneity and acute nature of the evidence are recognised. Across the studies that contrasted continuous and intermittent SSGs, intermittent formats were often associated with higher external intensity per unit of time (e.g., more high-intensity running or sprinting actions), whereas continuous formats tended to produce steadier internal load profiles (e.g., higher average HR or RPE). In some cases, shorter bouts increased the frequency of individual technical actions, while longer bouts challenged the execution of these actions under fatigue.

From a practical standpoint, such patterns suggest that continuous regimens might be useful when the goal is to expose players to sustained load and to technical execution under fatigue, while intermittent regimens may be useful for targeting repeated high-intensity efforts or for practicing technical skills with partial recovery. Alternating formats within a microcycle could therefore allow practitioners to emphasise complementary outcomes.

However, it is important to stress that these interpretations are context-dependent (e.g., age group, competitive level, game format, pitch size) and based on acute, cross-sectional evidence only. Technical and tactical outcomes remain especially scarce and inconsistently measured, making it premature to derive prescriptive guidelines. As such, the points summarised above should be viewed as indicative trends rather than firm recommendations, and further standardised, longitudinal research is required to confirm whether these acute responses translate into lasting training adaptations.