The main objectives of this review were to examine previous literature on (1) how different approaches to strength training for competitive swimmers can improve swimming performance and (2) which form of strength training resulted in the largest improvement in swimming performance. Collectively, almost all the experimental groups, and some of the control groups, showed a decrease in total swimming time and thereby gained a positive outcome of the training intervention. The results varied from a 7.5% performance increase  to a −1.45% performance decrease , with an average increase of 2.2% in the specific in-water training group, 2.5% in the non-specific dry-land strength training group and 2.6% in the dry-land swim-like training group. Furthermore, most of the studies were done in relation to the performance of the front crawl.
When assessing the results, there are important method-related inconsistencies that need to be considered. Firstly, there is a large age gap between the participants in the studies (13–24 years old), which leads to differences in competitive levels and training experiences that will influence the results. The highly skilled, older athlete with longer training experience has a smaller range of improvement than the younger more inexperienced athlete. Men were among the majority in the training groups (66.7%), and there was mixing of sexes in several of the groups. Some of the studies only had male participants [8, 21, 36, 37, 39, 41, 47,48,49]). Gourgoulis et al.  had young female participants and the rest of the studies had both male and female participants. Participants’ numbers ranged from 10  to 82 , with an average of around 16 participants. Statistically, a low number of participants reduce the statistical impact of the study, and the value of the study’s findings must be evaluated accordingly.
Furthermore, there was a wide span in the duration of the training interventions. The shortest intervention lasted for 3 weeks  and the longest for 16 weeks , with an average of 8 weeks. This is problematic in the sense that the participants in the longer interventions had more time to adapt to the training, which could result in a more accurate representation of the effect that type of strength training had on swimming performance.
Another inconsistency is the three studies that did not apply a swim-only approach to their control groups [8, 21, 42]. These control groups performed their usual dry-land hypertrophy training, while their experimental groups performed dry-land swim-like strength training , maximal strength training  and weight-assisted hypertrophy training . This makes it difficult to determine the effect of the training intervention as compared to that of a control group.
In-water Specific Resistance Training
Specific In-water Arm Strength Training
The interventions in this group were designed to increase arm strength through specific strength training in the water, and there were three eligible interventions. There were a hand paddle intervention , an arms-only intervention  and a POP device intervention . It is difficult to conclude that this type of training has a definite positive or negative effect on swimming performance. Firstly, there is limited available research, since there are only three studies in this category. The mean of the three arm-strength interventions showed an improvement of 1.7 ± 1.2% (Table 10). However, Barbosa et al.  did not find a significant effect for their experimental group in a 50 m fc with 0% change in performance and a trivial change (0.14) between-group ES. This study was conducted over the span of only 4 weeks. This allows very little time for adaption to training and could explain the lack of results. Konstantaki et al.  also showed no significant improvement pre- and posttest in 372 m fc and a small improvement between-group ES. In this intervention, the EG performed 20% of the weekly swimming training with arms-only. The lack of improvement could be due to the fact that this form of training alone is not enough to gain more strength in the arms than normal swimming does. Although swimming performance did not improve, a 186 m arms-only trial did. This supports the principle of specificity. The EG improved the parameter they practiced, but there were transfer issues to swimming performance. Toussaint and Vervoorn  conducted tests on 50 m, 100 m and 200 m fc, whereas the experimental group showed a significant gain in all distances. The CG also showed gains in performance but only in the 100 m test. The ES was small. The device used in this intervention is highly specific to swimming and could be the reason that the EG improved their swimming performance. The CG performed the same sprint training as the EG but only showed a gain in the 100 m test, which could indicate that the chosen method of sprint training is effective, but the sprint training with the device was even more effective.
Specific In-water Resistance Training
In this group of training interventions, the focus is specific in-water training with added resistance. This is a swim-specific way to gain strength and follows the principle of specificity that specifies that training should be as close as possible to the actual sport performance. The resistance is applied to the swimmers through resistance bands, parachutes or drag suits. The mean percentage for this group was 2.5 ± 1.9% (Table 10), and all studies, except Papoti et al. , had a positive effect on swimming performance. This tells us that this method is likely to result in a positive gain in swimming performance. A 2.5% change in performance is a considerable improvement in competitive swimming, but the SD shows that the variation of improvement differs greatly between the swimmers.
Assessing the drag suit and parachute trained experimental groups’ performances, there are large differences in results, despite the fact that these training methods arguably are very similar. In Dragunas et al. , the swimmers pulled a parachute behind them, and in Gourgoulis et al. , they wore a belt around their waist with pockets that filled with water when the swimmers swam, increasing the resistance. Dragunas et al.  had a 0.3% gain in 50 m fc performance, while Gourgoulis et al.  experienced a 3.2%, 5.1% and 7.5% gain in 50 m, 100 m and 200 m tests, respectively. The between-group ES was trivial in Dragunas et al. , and in the 50 m, 100 m and 200 m tests in Gourgoulis et al. , it was small to large (0.32, 0.49 and 0.89, respectively). The large variance in results could be due to the fact that the swimmers in Dragunas et al.  were 19–20 years old, and in Gourgoulis et al. , the swimmers were only girls that were 13–14 years old. The younger athletes have a large potential for improvement and possibly have greater use of this form of strength training than the older athletes that are already much stronger. Furthermore, the Gourgoulis et al.  intervention lasted for 11 weeks, where as Dragunas et al.  intervention lasted for only 5 weeks. The 11-week intervention allows for more time for adaption to training and could explain some of the reasons that this intervention had better results than the 5-week intervention.
For the resistance band trained experimental groups, the results were more consistent. In the resistance band trained groups, there were two methods of using the resistance band. Most studies had the participants swim out with the band to give resistance [29,30,31, 48]. The age of the participants ranged from 14 to 16 years old in all studies, and the mean gain in performance for the four interventions was about 2.0%. One study had a combined resisted-assisted method where the swimmers swam resisted one way and assisted the other way . This resulted in a 3.0% gain in performance. Girold et al.  had two experimental groups, one group swam resisted, and one group swam assisted, and then compared the two. The resisted group had a 2.0% gain in performance, which correlated with the other four resisted trained groups, while the assisted group had a 0.9% gain in performance and the lowest gain in performance for all the resistance band trained groups. These results indicate that if training with a resistance band is desired, a combined resisted-assisted method might be most successful. However, only one study had this approach, which makes the results tentative.
Specific In-water Leg Training
The arms are generally considered the main propulsive factor in swimming and are, therefore, often the focus when discussing strength training in swimming, even though the legs contain large muscles with great strength potential. Aspenes and Karlsen  speculate the legs in swimming are more of a stabilization factor to reduce drag rather than increase propulsion and swimming velocity. Gullstrand and Holmer  performed a correlation study with international level swimmers over a 5-year period and found that tethered leg kicking was not related to swimming performance. On the other hand, Schumann and Rønnestad  mentioned that a gain in leg strength could result in improvement in start and turn performance, which could result in an all-over gain in swimming performance. Only one study was eligible for this review. Konstantaki and Winter  executed a leg kicking study but found no significant change in a 400 m fc (-0.65%). The between-group ES was small (0.2). Arguably, a 0.65% gain in performance for an experienced swimmer is a positive effect, but considering the distance swam (400 m fc), this result is not of any real practical importance. Due to the limited availability of research, it was not possible to draw a definite conclusion of how an in-water leg training intervention could affect swimming performance. Compared to the in-water arm-strength training and the in-water resistance training, it seemingly would be beneficial to perform these methods of resistance training over the in-water leg training.
Dry-Land Swim-Like Resistance Training
This form of strength training is considered the most specific to swimming, when on dry land. It mimics the swimming performance, but it lacks specificity in the sense that the arms are isolated, the drag phase is longer than a swimming stroke in the water, and the distribution of the drag forces at various joint angles is not like in-water swimming . It is also worth considering that this form of training demands specialized equipment that may not be as accessible as a swimming pool, rubber bands or a strength training room.
The collective mean for these intervention groups was a 2.6 ± 1.9% enhancement in performance, but there were large differences in performance changes. The greatest change was in the Roberts et al.  study on 91.44 m fc, with a 5.0% increase in performance. However, this is probably not due to the swim bench training, as the CG also experienced large and almost the same gain in performance (5.1%) over the 10-week intervention. This could mean that other substantial factors have impacted the swimmers, as a 5% improvement is a huge enhancement in 91.44 m. Roberts et al.  speculated whether the improvements could be due to the fact that earlier in the season the main goal was to improve the biomechanics of the stroke and maximal VO2, while in the second part of the season, when the intervention took place, the focus shifted to a more high stroke turn over, anaerobic power and endurance, which are all important factors in a 91.44 m performance. The shift in focus obviously had a positive impact on the swimmer’s performance, but it is not certain that the swim bench training had an extra positive effect compared to the CG. Naczk et al.  used the same swim bench method as Roberts et al.  but found significant changes in the 50 m fc and 100 m butterfly (0.79% and 1.83%, respectively) in the EG only. However, Naczk et al.  also had limitations, as the duration of the intervention was relatively short (4 weeks). This provided little time to adapt to the training, making the findings uncertain. Naczk et al.  believed that some of the effects could be explained on the basis of placebo.
Sadowski et al.  and Sadowski et al.  used a device similar to the swim bench called a hydro-isokinetic ergometer. Sadowski et al.  performed a 6-week intervention and found a nonsignificant 1.2% gain in performance in the EG, while Sadowski et al.  performed a 12-week intervention and the EG had a significant 4.1% change in performance (as did the CG) (2.7%). The control group did not perform a swim-only method, but rather dry-land hypertrophy training. This made it difficult to ascertain the true effect of the ergometer vs. normal swimming practice, but it made it possible to compare swim-specific dry-land training and non-specific strength training. Both methods resulted in significant gains in performance, but the swim-specific method had greater improvements than traditional strength training. When comparing the two ergometer trained experimental groups, Sadowski et al.  showed the largest performance enhancement compared to Sadowski et al. , which was probably due to the duration of the interventions (12 weeks vs. 6 weeks).
Dry-land Non-specific Resistance Training
This type of training is non-specific to swimming, but it is widely used by swimmers due to the unstable nature of water, which demands a strong core for a purposively forward propulsion. The collective mean change in this group was 1.9 ± 0.8%, all measured in the 50 m fc (Table 10), which is a substantial improvement in such a short distance for experienced swimmers. However, Sawdon-Bea and Benson  indicated an insignificant change in performance for the EG of 1.7%, which was hard to explain. Some possible reasoning for the absence of a significant increase in performance probably lies in the fact that the participants were only experienced high school swimmers competing at a regional level, which could have affected the quality of core training they received due to variations in levels between the participants at this level. Furthermore, Sawdon-Bea and Benson  did not specify what kind of core exercises the participants executed. The exercises could lack an element of specificity that the other interventions had and therefore, was not always transferred to the swimming performance for each participant.
Traditional Resistance Training
Traditional resistance training is widely used in swimming and involves conventional gym-based strength training. In this review, traditional resistance training was divided into hypertrophy training, maximal strength training, plyometric training and a combined endurance and strength training regimen. The mean change in performance for these methods was 2.6 ± 1.5%, with only one study reporting a negative outcome in swimming performance . This was a hypertrophy training intervention with a focus on upper body strength. The EG in a study by Tanaka et al.  increased their weights by 25–35% over the span of the intervention but showed no gain in swimming performance or swim bench power. The lack of positive transfer could be due to a lack of specificity in the training. This may be an insufficient explanation for the decrease in performance, while the mean gain in performance in the hypertrophy trained groups was 2.6%. Trappe and Pearson  applied a weight-assisted hypertrophy strength training program for the EG, while the CG performed free-weight hypertrophy training. This made it problematic to investigate the differences between a combined hypertrophy and swimming training regimen and swimming training alone. Both the weight-assisted group and free-weight group gained significant change in the 365.8 m fc (around 3.8% for both groups) and had a trivial (0.03) between-group ES, which tells us that there is little difference between the two training methods.
It does not appear to be of importance whether the hypertrophy training was full body or upper body focused, as similar improvements were found after performing a full body strength training routine rather than an upper body focused one [21, 42, 48, 50]. This strays from the principle of specificity that says the upper body is the primary propulsion factor in swimming and that it seemingly would be most beneficial to perform upper body strength training. However, this is in line with the in-water resistance training groups where the added resistance trained group gained larger performance enhancements than the in-water arm strength only training group. This could mean that a full body focused resistance training regimen, regardless of whether it is in-water or on dry-land, is more beneficial to the transfer to swimming performance rather than just focusing on one part of the body (e.g., the arms).
In the maximal strength training interventions, the collective mean was 2.7 ± 0.8%, which states a possible likelihood of change in performance. Most studies conducted only the maximal strength training intervention and compared it with a control group, which gives a clear indication if the strength training has a positive effect or not. Only Aspenes et al.  conducted a study where they combined a 4 × 4 min endurance program and maximal strength training (a pull-down exercise designed to mimic the butterfly stroke). They investigated the 50 m, 100 m and 400 m freestyle, and the mean change in performance in the three distances was 1.3%. The only significant changes were found in the 400 m performance. The between-group ES never reached a significant level, except in the 100 m performance, with a small between-group ES (0.46). Therefore, in this study, it is difficult to predict whether the gain in the 400 m performance is due to the maximal strength training or to the endurance training, but it is suggested to be related to the strength portion of the program since the VO2max and work economy remained unchanged . Aspenes et al.  was the only study that tried to apply a specificity aspect to maximal strength training. This seemingly did not make a difference in the swimming performance, as the other maximal strength training groups had larger improvements in performance (2–3%). This may indicate that a general increase in strength is sufficient and preferred for an improved swimming performance.
Only one study investigated the effect of plyometric training on total swimming performance . Plyometric studies in swimming are often related to start-and-turn performance and Bishop et al.  showed positive effects in swimming performance after this kind of training. Potdevin et al.  showed a 3.1% and 4.7% change in the 50 m and 400 m fc, respectively, which is a considerable improvement. The CG also significantly improved their 400 m performance (1.1%), which makes it unclear if it is the strength training intervention or other factors that influenced the swimmer’s performance. Nevertheless, the gain in performance was larger in the EG, which tells us that maybe plyometric training had a positive effect. In the 50 m performance, only the EG improved their performance. This could be due to the shorter distance, where start performance plays a greater role in total performance than in the 400 m, and plyometrics has been shown to positively affect start performance . However, one study is not enough to conclude whether plyometric dry-land training has a positive or negative effect on swimming performance.
Comparison of Training Methods
It is an established fact that specificity in training is necessary for positive transfer to performance, but it is curious to note that all three groups had a mean gain in performance of 2–3%, which is a considerable improvement for competitive swimmers, regardless of what kind of strength training they performed. Regarding mean gain in performance, specific in-water training methods had a 2.2% mean gain, dry-land swim-like resistance training had a 2.6% mean gain, and dry-land non-specific strength training had a 2.4% mean gain. Thereby, the current literature demonstrates that various resistance training methods can positively impact swimming performance.
Dry-land swim-like resistance training showed the greatest change in performance, but this is also the group with the fewest studies and participants. Only one of four studies showed a statistically significant change in performance, which could be due to the lack of specificity in the movement of the swim bench. The non-specific dry-land training methods were used in 13 different studies. Three subgroups contained several interventions and made it possible to draw the following conclusions: (1) core training showed a 1.9% gain in performance, (2) hypertrophy training a 2.6% gain and (3) maximal strength training a 2.7% gain, which showed that all methods could positively affect swimming performance. Core training could be beneficial due to the nature of swimming, but it needs to be specific in the way that the core training on land is transferable to in-water swimming. Both hypertrophy and maximal strength training led to similar and considerable gains in swimming performance, which indicates that gain in muscle strength, even though the training is not specific to swimming, is transferable to swimming and has positive effects on performance. These methods showed substantially larger effects than core training, which might predict that hypertrophy or maximal strength training could be more useful to the swimmer than core training alone. Specific in-water training with 12 included studies had the least gain in performance. Nevertheless, the results showed that specific in-water strength training also leads to a probable gain in performance. The greatest all-over individual swimming performance improvements were found in this group. Within this group, the interventions with added resistance had greater gains in performance compared to the arms and legs focused interventions, which could be due to the principle of specificity. The act of swimming with a rubber band is more specific to swimming than swimming only using the arms.
When discussing the principle of specificity, it would be reasonable to conclude that the specific in-water training should lead to a greater gain in performance. There could be several reasons for this outcome, and due to the limited availability of literature, it is hard to make a definite conclusion. One reason may be that dry-land hypertrophy and maximal strength training leads to greater improvement in muscle strength than in-water resistance training and that might be what is needed to significantly increase swimming performance. It has been shown that younger athletes benefit from in-water resistance training [30, 31, 35], but for stronger and more experienced swimmers, in-water resistance does not necessarily result in increased muscle strength, which could be why dry-land strength training is more effective for improvement in swimming performance.
This review has three limitations. First, as there are limited studies in some of the categories it is still not possible to provide a definitive statement about which resistance training method is the most effective one to increase swimming performance. Secondly, it is possible that some studies were not found in the search process. Lastly, there are many other factors that could influence swimming performance over time which are possible confounding variables outside of the intervention programs since training is a multifactorial process.