The purpose of this paper was to conduct a meta-analysis that (1) quantified the effects of low (LF; 1 day week− 1), medium (MF; 2 days week− 1), or high (HF; ≥ 3 days week− 1) RT frequency on muscular strength per exercise; (2) examined the effects of different RT frequency on one repetition maximum (1RM) strength gain profiles (multi-joint exercises and single joint exercises); (3) examined the effects of different RT frequency on 1RM strength gain when RT volume is equated; and (4) examined the effects of different RT frequency on 1RM strength gains on upper and lower body.
This paper is the second systematic review that compares different RT frequencies and provides evidence from additional studies that investigates a graded dose-response relationship where strength gains are developed following increased training frequency. Furthermore, results from this meta-analysis highlight the need for further research exploring methods used in professional practice. Although this meta-analysis endeavoured to include research papers from high-quality sources, the number of suitable studies was small and there remained differences in design and control among included studies. This consequently produces issues that may influence data reliability including the low statistical power due to small pooled sample sizes.
Recommendations on Weekly Training Frequency
The existing evidence on the effect of weekly RT frequency has on strength development has been produced from limited and substantiated scientific evidence. Exercise physiology literature suggests that beginners train 2 to 3 days week− 1 and that more experienced subjects engage in more frequent training . The ACSM position stand  cites 16 RT studies that support their frequency recommendations for strength development; for untrained [17, 29, 53,54,55,56]; intermediate [16, 17, 49, 53, 57]; and well-trained subjects [38, 53, 58]. The position stand  recommends that novices (those with no RT experience or have not trained for several years) train the entire body 2 to 3 days week− 1. For intermediate subjects, a similar frequency of 2 to 3 days week− 1 total-body workouts or split routines (upper body/lower body) to provide a higher volume of exercise. The RT frequency of 4 to 5 days week− 1 for advanced weightlifters, powerlifters, and bodybuilders has been suggested for strength development.
A recent meta-analysis by Grgic et al.  compared different RT weekly frequencies (1, 2, 3, and ≥ 4 days week− 1) on muscular strength gains. The results of their analysis indicated a significant effect (p = 0.03) on muscular strength was achieved when weekly RT frequency was increased. The ES increased with each additional weekly RT session from 0.74, 0.82, 0.93, and 1.08 when training 1, 2, 3, and ≥ 4 days week− 1. Subgroup analysis for 1RM strength test on multi-joint exercise showed a significant effect (p ≤ 0.001), but not single-joint exercise (p = 0.324). Analysis of upper body revealed a significant effect of frequency (p = 0.004), but not the lower body (p = 0.07) on strength gains. A significant effect of training frequency was reported among young adults (p = 0.024) but not the middle or older aged adults (p = 0.093). In addition, subgroup analysis for sex identified a significant difference of RT frequency in females (p = 0.03), but not males (p = 0.19). However, when subgroup analysis was performed on volume-equated studies, no significant effect (p = 0.421) of RT training frequency on muscular strength gains was observed.
The results of this analysis (without accounting for training volume) cannot fully support the findings of Grgic et al.  regarding the contention that increased weekly training frequency is superior to lower weekly frequency. In this current review when combining multi-joint and isolation exercises, a similar strength gain relationship was observed with HF compared to LF. Analysis of upper and lower body pre- to post-strength was comparable when performing HF compared to LF. Upper body pre- to post-intervention strength gain was similar when MF was compared with LF. Lower body pre- to post-intervention strength gain was greater when HF was compared with MF but not statistically significant. The results of this analysis suggest that only negligible muscular strength increases are made with additional weekly RT sessions. The only findings in this analysis that support a significant relationship between RT frequency and strength gain were MF vs. HF in isolation and multi-joint exercises (ES 0.31; 95% CI 0.05–0.58; p = 0.02) and LF vs. HF for upper body (ES 0.48; 95% CI 0.20–0.76) p ≤ 0.01). However, readers should interpret these findings cautiously as limited study data were available to assess for a graded response relationship between medium and high frequency.
The differences that exist between Grgic et al.  and this review (excluding volume-equated analysis) could be due to confounding factors that may have influenced study outcome reliability. Grgic et al.  assessment of the consistency of effects across studies has not been included and is an essential part of the meta-analysis . Unless tests for heterogeneity are performed, it is difficult to determine the findings. The quantity I2 as in the current review was used to assess heterogeneity among subgroups , using only p values to decide which scale is more consistent with the data  is unsuitable because of the differing and limited number of studies. Likewise, the Benton et al.  study included within the Grgic et al.  was entered as an RT frequency of 2 vs. 3 days week− 1 instead of 3 vs. 4 days week− 1. This misrepresentation of study data leads to a detrimental effect on the 2, 3, and ≥ 4 days week− 1 RT frequency groups that consequently effects the accuracy of pooled mean ES results. It could be assumed that this accounts for the variances between frequency groups and concerning the two papers strength outcomes.
Rhea et al.  conducted a meta-analysis [in part] that sought to quantify the optimum dose response for trained and untrained subjects along the continuum of weekly frequency, volume, and training intensities. Rhea et al.  provided evidence that may support the contention that increased weekly training frequency is superior to that of single training sessions per muscle group. The researchers reported that the ES for training frequency was different by training status. Rhea et al.  stated that the ES increased for untrained groups as RT frequency increased up to 3 days week− 1. However, the trained subject’s ES elicited the most significant strength increases with a weekly training frequency of 2 days week− 1 . The RT design for the trained group had increased training volume that may have been too challenging for the untrained subjects. Research by Hoffman et al.  and Stowers et al.  suggest that trained athletes are possibly closer to their strength potential and that higher training frequencies may evoke more significant strength gains. Moreover, Hoffman et al.  and Stowers et al.  suggest that smaller muscles produce smaller observed strength gains, and this may require the subjects to have more stimulus or more extended observations before reporting statistically significant differences. Evidence from this analysis did not detect significant variances in strength gain when LF was compared to HF in isolation-only exercises (ES − 0.10; 95% CI − 0.43–0.23; p = 0.56). However, limited study data were available to assess for a graded response relationship between lower and high frequency.
Considerations Towards Weekly Training Frequency and Volume-Equated Studies
Centred upon the available body of evidence from two meta-analyses [19, 53] and other recent studies [61,62,63,64], it may be suggested that RT volume is a causal factor to increase muscular strength. Depending upon the subjects training status, additional RT training frequency could attribute to changes in muscular strength for untrained subjects due to increased weekly training volume. Examination of pre- to post-strength gain from volume-equated studies in this analysis was comparable when LF was compared to HF. Therefore, not equating for weekly RT training volume in studies that compare strength gains might be erroneous. Limited extrapolation can be made of the effects on muscle strength due to higher RT frequency or increased weekly RT volume. This is supported by subgroup analysis of volume-equated studies in Grgic et al. , and this analysis which did not show a significant effect of RT frequency on pre- to post changes on muscular strength. The ES was similar across lower and higher RT frequency strength outcomes.
A recent study by Colquhoun et al.  suggests that additional RT frequency does not lead to further strength improvements when volume and intensity are equated. Male subjects were randomly assigned to either 3 or 6 days week− 1 training intervention. Pre-and post-baseline strength measurements after 6 weeks indicated that no significant differences between 3 and 6 days week−1. This raises several questions concerning the significance of weekly RT volume rather than RT frequency. A recent analysis that we conducted on weekly set volume  suggests that there is a graded dose-response relationship between RT volume and muscular strength gains. We concluded that lower weekly set training produced the smallest pre- to post-training strength differences when compared to medium or higher weekly set training. Further support regarding the importance of weekly training volume on muscle hypertrophy is provided by the meta-analysis of Schoenfeld et al. . From the 15 included studies, a significant effect was reported in muscle size due to increased weekly RT volume. The ES difference between lower and higher volumes equated to a difference of 3.9% strength change. Figueiredo et al.  state that volume is the most modifiable variable that has the most evidenced-based response with significant physiological effects on muscle. Future research is required from study designs that equate for RT weekly volume to clarify the effect of RT frequency on strength.
Strengths and Limitations
This meta-analysis has several strengths that separate it from other previous analyses of training frequency. This analysis attempted to apply more robust criteria to try to control potential confounding variables when comparing the effects of weekly training frequency on strength outcomes. Our intent to create an evidence-based dose-response curve of frequency to strength gain was subverted and resulted in a high- vs. low-frequency comparison for some strength measures. This meta-analysis also considered the possible effects of different sections of the body and the impact it has on strength outcomes on the impact of LF or HF weekly training frequency. The design of this study also differed from others, as it did not cluster outcomes. Instead, data were combined across strength measures to improve external validity. Within our design, we considered and included a multi-level model as a strategy for testing heterogeneity across included studies.
As with previous meta-analytic studies, there were limitations driven by the shortcoming of primary data sources. This present meta-analysis attempted to include relevant and frequently cited research data from high-quality sources, the number of studies was small, and variation existed in the design and control of the included studies. Although every effort was made to include research papers from high-quality sources, the number of suitable studies was limited, and the research designs and control among studies were different. Unfortunately, even when controlling for confounding factors, the low number of studies and sample sizes used in this meta-analysis may exert an effect on estimates of ES. The authors have attempted to ensure that all included studies were appropriate due to the initial screening process. This created difficulty in summarising and interpreting study data.
The validity and utility of this analysis should be evaluated with caution as there are limitations due to the inclusion of combined subject’s characteristics (for example, male-female or trained-untrained). This sampling of mixed gender groups, use of extensive age ranges, use of multiple and different measurements, and the use of various training methods has resulted in a moderately large body of evidence that may be deemed unreliable and not provide answers to strength gain questions for individuals or collectively for groups.
The limitations of previous research by default extend to the present meta-analysis deriving data from that research. Two of the 12 included studies used a randomised control design [11, 52]. The other 10 [17, 22, 44,45,46,47,48,49,50,51,52] did not use a control group. They used a repeated measures design with baseline measure serving as the control, although baseline measures were not uniformly implemented across those studies. The finding of the present analysis suggests that researchers should be cautious when performing mixed-model meta-analyses (mixed gender subject groups and diverse training groups), as this could limit data analysis and produce spurious conclusions. While studies that combine subjects with differing characteristics can provide useful data, there are at the same time limitations in applicability and relevance. For example, combining males and females in a subject pool or including both trained and untrained in a subject pool (or not fully describing training state) will limit the extent to which the findings may be generalised to either population. When analysing strength gain per exercise, it creates confounding aspects that are difficult to control. For example, different exercises that target the biceps and then measure the strength of a lat pulldown will have a direct effect on the strength measurement. Such designs, common in frequency research, are not as reliable as single model data methods.
It is often stated that the design of RT programs is multifaceted, requiring manipulation of several training variables that interrelate with each other. One of those variables is weekly RT frequency. This is not unique to strength training as the FIT approach to programming, Frequency-intensity-time as variables is commonly taught as axiomatically within physical education and exercise science curricula. However, with further investigation, one finds that any attempt to establish and define an optimal training frequency is undermined by conflicting findings and a lack of clear methodological clarity and consistency from previous study protocols. Those issues with methodological clarity create inadequate data estimations from published studies when performing meta-analyses.
Such equivocality creates conditions under which it is difficult to establish any definitive conclusions. Future investigations and research should be as task-specific as possible and with consideration of training status on test validity . As subjects perform pre- to post-1RM measurements, considerations should be made, as this is a task-specific skill that could incorrectly represent relative increases in strength . Attention should also be made concerning training specific tests and relationships between training frequency and improved 1RM performance. As previously mentioned, there is limited primary data with which to develop an evidence-based consensus regarding the best weekly RT frequency to produce strength gain. More considerable attention needs to be placed upon designing and conducting larger studies using homogenous sample pools (similar biological characteristics and training histories). Increased homogeneity and larger sample sizes would improve primary research but would also strengthen meta-analyses. Replication of studies would also be beneficial and would allow the data and findings to be corroborated.
A better body of research evidence from more studies would also have a profound effect on meta-analyses. Performing meta-analyses on RT variables that were not controlled or inadequately controlled and were conducted in heterogeneous samples is problematic. This is because such weak study designs lead to the exclusion of a significant of extant research publications. This, therefore, leads to variability in methods and results reporting among the best research on the topic remains varied and un-replicated, conclusions from their pooled analysis, while stronger than individual report, remain weak. A significant non-experimental finding of this project was that there was very little experimental evidence of any quality or consistency published related to RT frequency.