Our review is the first to comprehensively collate research reporting the external and internal loads experienced during training and games in female basketball players. Despite 32 studies being conducted on this topic, surprisingly few load variables have been measured following consistent methodologies across studies. The non-standardized measurement of external and internal load variables across studies prevented the ability to draw definitive conclusions regarding the typical training and game loads experienced by female basketball players according to playing level and position for most variables. From a practical perspective, inconsistencies in the literature regarding the seasonal phase monitored, minimum exposure time set for including player data, approach to measure session duration, approach to measure session intensity, and duration of monitoring periods make it difficult for basketball coaches and researchers to select appropriate load variables and follow uniform procedures when monitoring female basketball players. To address this issue, we provide recommendations to enhance the methodological rigor and promote greater consistency in approaches adopted across future studies investigating external and internal loads in female basketball players.
External and Internal Loads During Training
Individual Training Sessions
Weighted means for loads experienced during individual training sessions in female basketball players could only be calculated for average net force in professional players and sRPE in club and representative players. Specifically, average net force ranged from 272 ± NP N  to 293 ± 40 N in professional players  with a weighted mean of 281 ± NP N [40, 41]. In this regard, the highest average net force value for individual training sessions was indicative of longitudinal monitoring across 18 training sessions , while the lowest value reported for individual training sessions was indicative of longitudinal monitoring across 54 training sessions . Analyzing fewer total training sessions may skew results as acute monitoring periods likely misrepresent the average net force experienced across the wider season due to factors that could allow coaches to administer increased training loads across acute timeframes (e.g., more days between games, less or no travel for games, fewer games played). Furthermore, while data from these studies [40, 41] were collected during different basketball seasons, the outcomes reported were indicative of the same professional basketball team. Consequently, the inclusion of new players, progression of physical (e.g., lean muscle mass and percentage body fat) or physiological (e.g., speed and anaerobic capacity) attributes in players, and potential changes in training approaches or coaching staff between seasons may have contributed to the variation in average net force reported across studies.
Multiple studies reported sRPE during individual training sessions in club and representative players. In this regard, 1 of the 2 studies  investigating sRPE in collegiate players failed to report a measure of intensity (i.e., RPE), while both studies neglected to report training duration [44, 51]. Furthermore, 1 of the 2 studies  investigating sRPE in collegiate players failed to specify the RPE scale used, preventing us from calculating a weighted mean. Accordingly, we recommend future studies should aim to clearly report the constituent data comprising sRPE values (i.e., RPE scores and session durations) as well as identify the specific RPE scale used to allow for meaningful comparisons in sRPE data across studies examining female basketball players.
Given the amount of published research exploring load monitoring in female basketball players, the fact that average net force and sRPE were the only load variables reported during individual training sessions across multiple studies highlights a lack of attention given to understanding how training is prescribed at the session level as opposed to longer periods (e.g., weekly, seasonal phase). Furthermore, based on the available data, it is unclear how the loads experienced during individual training sessions vary between female players competing at different playing levels or occupying different playing positions. We recommend future studies quantifying weekly external and internal training load to report the load experienced during individual training sessions to allow basketball coaches to better understand how training volume and intensity are altered between weekly microcycles across the season.
Total Daily Training Load
We were only able to calculate a weighted mean for loads accumulated across all training sessions completed in a day in female basketball players for total daily PL and internal TRIMP in representative players. In this regard, total daily training PL ranged from 706 ± 295 AU in U20 representative players  to 816 ± 333 AU in U18 representative players  with a weighted mean of 761 ± 314 AU  across age groups, while internal TRIMP ranged from 215 ± 109 AU in U20 representative players  to 305 ± 172 AU in U18 representative players  with a weighted mean of 260 ± 141 AU  across age groups. Given the available total daily training PL and internal TRIMP data were reported in the same study for different player samples (i.e., U18 and U20 players) during intensive training camps, the variance in daily load is likely explained by the different training configurations prescribed for each age group rather than methodological inconsistencies. In this way, U20 representative players completed fewer daily training sessions than U18 representative players during the training camp (U18: 14 out of 21 days had 2 training sessions; U20: 8 out of 18 days had 2 training sessions ), reducing their activity exposure to lower the average accumulated daily loads experienced.
Weekly Training Load and Weekly Training and Game Load
Although multiple studies reported weekly training sRPE in professional female basketball players, differences in the seasonal phase monitored, RPE scale used, and monitoring period duration prevented weighted means from being calculated across studies [47, 53]. Specifically, Nunes et al.  observed 19 professional basketball players from the Brazilian National Team during a 12-week preparatory training camp, while Paulauskas et al.  examined 29 professional basketball players from the first division Lithuanian Women’s Basketball League during a 24-week in-season period. In this regard, preparation periods typically involve longer and/or more frequent training sessions at higher intensities (i.e., overloading) than the in-season to promote positive adaptations in preparation for competition . In turn, lower training loads are typically encountered during the in-season compared to preparatory training periods among basketball teams to optimize player readiness for games . Consequently, the weekly loads experienced by female basketball players are likely dependent on the seasonal phase monitored, which should be clearly described in future studies and considered when interpreting reported data. Additionally, 1 of the 2 studies  investigating weekly training sRPE in professional players failed to clearly identify the RPE scale used. Given the absolute sRPE value derived when monitoring loads is dependent on the RPE scale used , calculating a weighted mean across studies not clearly specifying the RPE scale adopted might yield misleading findings.
Weekly training and game sRPE was only reported across multiple studies in club players ranging from 879 ± 140 AU  to 1215 ± NP AU  with a weighted mean of 1161 ± NP AU [48, 50]. The variation in weekly training and game sRPE reported is likely explained by the monitoring periods utilized across studies. Specifically, Ghali et al.  collected data across a 1-week period at some point in the season that was not identified, while Otaegi and Los Arcos  collected data across a 9-week in-season period. The longer monitoring period utilized by Otaegi and Los Arcos  likely encompassed week-to-week fluctuations in training and game sRPE experienced by players whereby training was likely adjusted dependent upon in-season factors, such as game scheduling and travel requirements. In turn, the shorter monitoring period utilized by Ghali et al.  was likely not representative of the typical weekly training and game loads encountered across the entire season given week-to-week fluctuations in sRPE as high as 47% have been reported across the in-season phase in professional female basketball players . As such, future basketball research should aim to maximize the monitoring period duration to best understand the typical weekly training and game loads imposed on female players.
The lack of studies reporting weekly training and game loads in semi-professional and professional players is surprising as basketball teams competing at these levels likely possess more resources (e.g., finances, staff expertise) than teams competing at lower levels to implement comprehensive player monitoring systems. Furthermore, load data are essential to permit evidence-based decisions that optimize the training and game stimuli encountered, readiness for games, and risk of maladaptive responses in players competing in semi-professional and professional leagues given the arduous demands they face [39, 49]. The deficiency in studies reporting weekly training or weekly training and game loads in semi-professional and professional female basketball players currently limits the ability to comprehensively compare data across playing levels, which can be used in benchmarking processes when transitioning players to higher playing levels.
External and Internal Loads During Games
Despite multiple studies reporting activity distances, frequencies, and durations in female basketball players across different playing levels and positions, weighted means could not be calculated due to several methodological variations across studies. First, this review identified 9 studies reporting movement frequency, duration, and distance covered during basketball games using different technological approaches (video-based TMA, microsensors, and LPS) along with different software packages (LabVIEW, Dartfish, sPRO, SVIVO, Openfield, WIMU, Dynamic Image Analysis System, and LINCE multiplatform analysis). While the use of various technologies across studies is inevitable due to prohibitive factors such as cost and the long-term availability of equipment, the use of various software packages likely introduces variation in the acquired data given undisclosed proprietary algorithms and filtering processes are used in some packages. Second, the number (i.e., 1–4), brand (i.e., Sony, Basler, JVC, DKH, or not reported), positioning (e.g., placement around court, distance from court, height above court), and recording frequency (i.e., 7.5 Hz, 25 Hz, 30 Hz, or not reported) of cameras used for video-based TMA varied between studies. These camera-related variations across studies likely impact the data given the accuracy of vision-based systems is affected by the distances between cameras and players, camera angles, and lens type in the cameras. Third, studies categorized movement and intensities using various methods (irrespective of monitoring technology), including subjective movement categories and intensities identified using frame-by-frame playback of video [29,30,31, 66], objective speed zones with no justification [33, 34, 60], and objective speed zones  based on research examining other court-based team sports . The use of various methods to categorize activity movement and intensity likely impacted the reported outcomes as the criteria used to define a given activity (e.g., sprint) were inherently inconsistent across studies. For example, one study  categorized running activity as multidirectional movement performed at 3.1–7 m·s−1, whereas two studies categorized sprinting activity as forwards or backward movement performed at > 4 m∙s−1  or > 5.8 m∙s−1 . Consequently, methodological inconsistencies between studies impeded the ability to definitively determine the typical activity demands experienced during female basketball games according to playing level and playing position.
We were only able to draw conclusions for BLa given it was the only variable reported across multiple studies. BLa is used as an indicator of energy re-synthesis from rapid glycolysis [29, 32]. In turn, BLa ranged from 3.7 ± 1.4 mmol·L−1 in semi-professional players  to 5.3 ± 1.9 mmol·L−1 in professional players  during games. The BLa values reported highlight the utilization of the rapid glycolytic energy pathway in executing game activities in female basketball players [29, 56]. As such, implementation of anaerobic conditioning drills incorporating prolonged and repeated high-intensity actions  is essential to improve tolerance of high BLa and enhance lactate threshold markers in female players. In this regard, aerobic conditioning is also critical to maximize lactate clearance and improve phosphocreatine regeneration during recovery periods between repeated high-intensity activities across games . Moreover, given multiple studies [32, 56] reported BLa in female basketball players during games according to playing position, we were able to calculate and compare weighted means for backcourt and frontcourt players. In this regard, a higher BLa was apparent in backcourt players compared to frontcourt players (5.2 ± 1.9 mmol·L−1 vs. 4.4 ± 1.8 mmol·L−1) [32, 56]. These position-specific variations in BLa might be explained by the strategic roles typically performed in each position during games. Specifically, backcourt players typically undertake frequent intense cutting movements to create space for open perimeter shots and defend opposing perimeter players cutting to receive the ball . Moreover, backcourt players are more likely to be involved in fast breaks as they initiate steals  or leak out when transitioning into offense as well as pursue opposing backcourt players when transitioning to defense. These intense movements performed frequently across games by backcourt players likely increase the reliance on rapid glycolysis for energy re-synthesis [72, 73] compared to frontcourt players who are typically positioned closer to the basket on offense and defense.
While multiple studies reported the absolute and relative HR of club and collegiate female basketball players as well as absolute and relative HR according to playing position during games, some key methodological variations across studies impeded the ability to calculate weighted means and draw definitive conclusions. First, ‘total time’ was inconsistently defined across studies, with studies defining ‘total time’ as the time during which the player was on the court including stoppages in play but not time-outs or breaks, including all stoppages in play (i.e., free-throws, out-of-bounds, and time-outs) but not breaks [29, 37, 43], or including all breaks and stoppages in play . Given rest periods between quarters and halves as well as during stoppages in play enable extra opportunities for recovery and reductions in HR, the inconsistent inclusion or exclusion of breaks and stoppages in play would have altered the outcomes reported across studies. Second, HRpeak was determined using various methods, including peak responses taken during an incremental treadmill test [29, 43, 59], peak responses taken during basketball training sessions , and peak responses taken during a 20-m shuttle run , or the method to determine HRpeak was not reported . Third, playing time criteria for including HR data from players were not specified [35, 37, 56, 58] or varied across studies with some studies using player data regardless of total playing time [35, 37, 59], if players accumulated ≥ 3 min of live playing time in any given quarter and ≥ 10 min of live playing time for the entire game , or if players accumulated ≥ 25 min of live playing time for the entire game . The use of different playing time criteria for data inclusion likely impacted the reported outcomes as shorter playing times are expected to elicit higher HR values during live game time but lower HR values during total game time compared to longer playing times. For example, during live game time, short spurts of activity are likely to produce rapid spikes in HR as a result of an increased oxygen deficit, while the inclusion of stoppages such as time-outs, out-of-bounds, and free-throws is likely to decrease the HR response during total game time due to increased recovery opportunities.
Limitations and Future Directions
Our review provides important information for basketball coaches and performance staff regarding the external and internal loads experienced during training and games in female basketball players; however, there are limitations that must be considered when applying the reported findings. On a positive note, the limitations encountered in conducting our review have brought much needed attention to the methodological inconsistencies across published research examining load monitoring in female basketball players, permitting us to develop recommendations aimed at improving the quality of future research in the field.
First, given the limited number of studies reporting external and internal loads in players competing in the same basketball league, we were unable to aggregate data according to basketball league. The game rules and competition format (e.g., game scheduling, game durations) are inconsistent across many basketball leagues, which may impact the external and internal game loads experienced by players and should be taken into account when interpreting the data presented.
Second, defining the type of players involved in studies is critical for understanding differences in external and internal loads between playing levels and playing positions, which is essential to develop training targets for basketball coaches. However, descriptors used to classify playing level and playing position were inconsistent across the included studies, which limited the ability to compare findings between studies. For example, the term ‘elite’ was used to describe several playing samples ranging from youth players in U14 club teams, collegiate players, and professional players. Regarding playing position, some studies categorized players into two playing positions as either frontcourt and backcourt [32, 40, 61] or guards and posts , while other studies categorized players into three (i.e., guards, forwards, and centers [33, 34, 43, 55, 56, 58,59,60]) or five (i.e., point guard, shooting guard, small forward, power forward, and centers [31, 37]) playing positions, but with different categorical criteria for each position. Therefore, to allow for comparisons between studies, playing level data were recategorized from lowest to highest as follows: club, high-school, collegiate, representative (trained athletes selected into a representative team), semi-professional (some players are full-time/contracted athletes), or professional (all players are full-time, contracted athletes), while positional data were recategorized into backcourt and frontcourt. Future research should seek to establish a consensus regarding the categorization of playing level and playing position in basketball research to better allow for comparisons between studies.
Third, external loads reported in our review were derived from various technologies, including video-based TMA, LPS, and microsensors (containing triaxial accelerometers, gyroscopes, magnetometers or a combination of these instruments). In this regard, the criteria (i.e., speed or intensity zones) used to distinguish between movement intensities and the formulae or algorithm used to calculate external load variables (i.e., Catapult PL vs. player load) were inconsistent across studies. Using various criteria (e.g., speed cut points) to distinguish between movements performed during training and games is likely to over- or under-estimate the external intensities being performed and prohibit meaningful comparisons in findings across studies. Consequently, expert consensus should be sought to establish cut points for basketball-specific speed or intensity zones with different approaches to monitor external load to allow for consistent and accurate classification of movements or intensities in future basketball research.
Fourth, training and game durations were determined inconsistently across studies, with some studies not specifying the methods adopted to measure session duration. This limitation should be considered when interpreting the data reported in our review. In turn, future basketball research should be transparent and detailed in describing the procedures used to measure training and game duration, with separate reporting of warm-up and cool-down components alongside other session components being advocated .
Finally, data collection was predominantly reported across acute periods in the included studies (12 ± 9 weeks). While the duration of data collection may vary based on the specified research aims across studies, the acute time periods used in most studies may produce skewed results due to the impact of factors that can directly influence training prescription and game demands such as game scheduling [75, 76]. Furthermore, most studies (67%) monitored players during the in-season phase only. The use of a single seasonal phase limits the applicability of the reported outcomes in practice as training load fluctuates across seasonal phases due to changes in training approaches and the physiological capacities of players [77, 78]. As such, we recommend future research to examine longer monitoring periods as well as different seasonal phases to gain a comprehensive understanding of the external and internal loads experienced in female basketball players during training and games across the annual plan.