What are cluster sets?
Cluster sets are strength training sets that involve intra-set rest periods. Curiously, the use of intra-set rest periods has been extensively researched in the literature, with many long-term training studies being available (as well as several acute investigations). Importantly, however, although cluster sets can be performed in two completely different ways, only one way has been researched in detail.
Like any other strength training set configuration, cluster sets can be performed either  reaching failure, or  without reaching failure. To date, the vast majority of the literature has explored cluster sets that do not involve reaching failure. This has been done so that the total number of reps performed in each set can be matched with a conventional set configuration for comparison purposes. For example, one training group might perform a cluster set training program comprised of 3 sets of 10RM (2 clusters of 5 reps) with a 30-second intra-set rest period between clusters, while the comparison training group might do a conventional set configuration that involves 3 sets of 10RM. Clearly, the conventional set configuration will involve training to failure on all 3 sets, while the cluster set training program will not, since the intra-set rest period allows time to recover from the accumulated fatigue in the first 5 reps of the set.
Do cluster sets work for enhancing improvements in athletic performance?
When the total number of sets with a given weight is matched (such that using intra-set rest periods involves training with less fatigue and therefore further from failure), the literature has shown that training with intra-set rest periods allows similar gains in maximum strength but greater gains in the various measures that underpin speed, such as power output, high-velocity strength, rate of force development, and maximum theoretical velocity on the force-velocity profile.
In this way, cluster set training produces a very similar set of effects to velocity-based training, which also minimizes the amount of fatigue that is experienced during a strength training set and which also produces similar gains in maximum strength to conventional strength training but superior improvements in measures that underpin speed. This is because training while minimizing fatigue seems to preserve type IIX fiber proportion, which is a key contributor to high-velocity force production. Indeed, the maintenance of type IIX fiber proportion has been directly linked to superior gains in speed and high-velocity strength measures after velocity-based training.
Do cluster sets work for enhancing hypertrophy?
When the total number of sets with a given weight is matched (such that using intra-set rest periods involves training with less fatigue and therefore further from failure), the literature has shown that training with intra-set rest periods does not enhance hypertrophy. Indeed, we should not expect it to produce any superior muscle growth, because the number of stimulating reps that are performed is either similar or slightly less (depending on the exact way in which the clusters are performed).
Nevertheless, as explained above, cluster sets also provide the option of performing additional reps per set. When performed in this way, cluster set training would more closely resemble rest pause training, insofar as the total accumulated fatigue would be similar to traditional set configurations at any given point (such that the superior effects on speed would be lost), while the duration of time spent exerting a high effort would be increased.
This is the type of cluster training that I aim to discuss in this article.
The key question is whether an extended duration of time spent exerting a high level of effort corresponds to additional stimulating reps or not (and so produces more hypertrophy). Despite not being able to answer the question in a conclusive way as yet, we can look carefully at the underlying physiology such that we can make an informed assessment of whether cluster sets have the potential to produce superior muscle growth.
How can we gauge the amount of hypertrophy that a strength training set might produce?
When comparing the effects of set configurations, we need some kind of model that allows us to quantify the stimulus that is provided by the reps that are performed. In previous articles, I have suggested that we can use a “stimulating reps” model in which reps that are performed within five or six reps of failure can be expected to cause hypertrophy, while all other reps cannot. This model is based on the underlying physiology of  motor unit recruitment, and  muscle fiber mechanical tension due to the force-velocity relationship. This model allows us to explain how strength training with the same number of sets of light loads (between 16RM and 30RM) to muscular failure produces essentially the same hypertrophy as strength training with moderate loads (between 6RM and 15RM), despite involving a lot more volume load (sets x reps x weight).
The strength of this model is that it assumes the same mechanism of effect during the two types of training (light loads and moderate loads), which explains why the hypertrophy is nearly always exactly the same. In contrast, models that propose different mechanisms of effect for the two types of training (such as metabolic stress during light load training and mechanical tension during moderate load training) are weak, because it is very hard to believe that the exact same hypertrophy will be produced in almost every single study if completely different mechanisms are involved.
Nevertheless, the stimulating reps model only allows us to compare the effects of different strength training sets if no other factors are changed (such as the nature of the mechanical tension that is produced or the fatigue that is being experienced). Indeed, this is immediately apparent if we compare the effects of two different exercise variations, one that involves training with a long maximum muscle length (such as a full squat) and the other that involves training with a short maximum muscle length (such as a partial squat). Training with a full squat will produce more hypertrophy not because of a great number of stimulating reps per set, but because the working muscle fibers experience higher levels of mechanical tension due to reaching longer sarcomere lengths on the length-tension relationship.
In the same way as the length-tension relationship can have a positive effect on the hypertrophy caused by a strength training set, fatigue mechanisms can have a negative effect. For example, when multiple strength training sets to failure are performed with short rest periods, this produces less hypertrophy compared to when the same number of sets to failure are performed with long rest periods. This occurs due to the negative effects of certain types of fatigue accumulating in the early sets on the adaptations stimulated by the later sets. Consequently, to understand how cluster sets might be useful for hypertrophy, we need to clarify exactly how fatigue works during a strength training set, and how that might affect the ability of the reps during cluster sets to stimulate muscle growth.
How does fatigue differ between cluster sets and conventional strength training sets? (the different fatigue mechanisms)
Fatigue is a key factor that affects the ability of a strength training set to stimulate adaptations. While fatigue has traditionally been assumed to contribute positively to the adaptive process, it actually has a negative impact, although this depends on the exact fatigue mechanism that is in operation. Several different fatigue mechanisms arise during strength training sets, including supraspinal central nervous system (CNS) fatigue, spinal CNS fatigue, calcium ion accumulation inside the working muscle fibers, and metabolite accumulation inside the working muscle fibers.
In general, supraspinal and spinal CNS fatigue have a negative effect on the ability to produce strength training adaptations (including both increases in the ability to recruit high-threshold motor units and hypertrophy) because they reduce the magnitude of the central motor command that reaches the working muscle to produce motor unit recruitment. When CNS fatigues are present, full motor unit recruitment is not possible, even when a maximal effort is being exerted (yes, it really does not matter how motivated or experienced the lifter, the presence of CNS fatigues will stop a maximal effort from translating into full motor unit recruitment. Yes, really, all the time. No, that particular situation or individual is not different).
Similarly, calcium ion accumulation (which causes a number of different types of fatigue mechanism to occur inside the working muscle fibers) also has a negative effect, but not because it affects motor unit recruitment. Rather, the presence of calcium ion-based fatigue mechanisms causes a reduction in muscle fiber mechanical tension, which then impairs the ability of the working muscle fiber to stimulate hypertrophy. In contrast, metabolite accumulation (specifically, acidosis) has the potential to produce a positive effect, because it primarily reduces muscle fiber shortening velocity and not muscle fiber mechanical tension. In this way, the arrival of acidosis inside a muscle fiber does not reduce its ability to stimulate muscle fiber growth, even though it still reduces exercise performance and therefore cues the lifter to increase effort levels and thereby recruit more motor units.
How does fatigue differ between cluster sets and conventional strength training sets? (how fatigue works during a set)
During strength training sets, fatigue accumulates. Specifically, various fatigue mechanisms take effect. Importantly, which fatigue mechanisms are present has a huge impact on the adaptations that are present. In the truly hypothetical scenario in which only metabolite-related fatigue were to be present, then a great many stimulating reps could be achieved. In contrast, in the similarly hypothetical scenario that only CNS fatigues were present, fewer than five stimulating reps would be achievable. While such extremes are not possible physiologically, some variance does occur from one type of training to another (which is why all this is relevant for understanding cluster sets). However, to understand how such variances might occur and how they might affect hypertrophy, we first need to be clear about how fatigue develops during conventional strength training.
During conventional strength training sets to failure, metabolite-related fatigue is the first fatigue mechanism to appear, and the resulting acidosis reduces maximum muscle fiber shortening velocity. When lifting weights with a self-selected tempo, lifters find themselves needing to increase their effort levels in order to maintain the same bar speed. This increase in perceived effort is caused by an increase in the central motor command that is sent to the muscle to increase the level of motor unit recruitment and therefore to increase the number of activated muscle fibers (the central motor command sends a second signal, called the “corollary discharge,” to another part of the brain, which we detect as an increase in the perceived effort level).
When the larger central motor command reaches the muscle, it increases the number of recruited motor units, and this increases the number of activated muscle fibers. These additional muscle fibers compensate for the reduction in muscle fiber shortening speed that has affected the previously-activated muscle fibers. In this way, acidosis creates a situation in which the lifter must increase their central motor command if they want to continue lifting a weight with the same tempo (but without greatly reducing the level of mechanical tension that the muscle fibers produce). And, importantly, by increasing the level of the central motor command, the lifters necessarily also experience an increase in the perception of effort during the rep being performed, by means of the increased corollary discharge.
Later, towards the end of the conventional strength training set to failure, calcium ion-related fatigue mechanisms appear, and these reduce maximum muscle fiber force. From the point of view of the brain, there is no difference between these two types of peripheral fatigue, it is only the effect inside the working muscle fibers that differs. Consequently, when lifting weights with a self-selected tempo, lifters again find themselves needing to increase their effort levels in order to maintain the same bar speed and complete the set. This increase in perceived effort is once again caused by an increase in the central motor command, and this again increases the perception of effort by means of the corollary discharge.
During the entire conventional strength training set to failure, both supraspinal and spinal CNS fatigues develop. Spinal CNS fatigue seems to develop linearly as a result of the duration of time that the motor neurons spend firing, and therefore it is greater when training with light loads to failure than when training with moderate loads. In contrast, supraspinal CNS fatigue seems to develop in response to the presence of afferent feedback from the working muscles, which we perceive as burning and fatiguing sensations. The presence of these perceived sensations increases our perception of effort. Since we have a maximum tolerable level of perceived effort, the presence of this afferent feedback prevents us from increasing the magnitude of central motor command to its normal maximal level, and this inhibits the brain from recruiting as many motor units as it would normally be able to achieve. During strength training, it is usually the accumulation of a combination of metabolites towards the end of a strength training set that stimulates the metaboreceptors to produce the necessary afferent feedback that creates these perceived sensations. For this reason, supraspinal CNS fatigue likely occurs increases exponentially rather than linearly.
For the sake of completeness, it’s worth pointing out that muscular failure is therefore a voluntary cessation of exercise in response to a predicted level of effort that would be intolerable. We reach muscular failure once we arrive at the maximum tolerable level of effort during the set. Once we reach this maximum tolerable level of effort, we are unable to increase the level of motor unit recruitment any further, and this means we cannot do anything to increase force production to compensate for the various fatigue mechanisms. This is why fatigue researchers prefer the term “task failure” rather than “muscular failure” because muscular failure implies that the muscle has somehow failed, but the reality is that any failure to continue performing the task always occurs inside the brain.
How does fatigue differ between cluster sets and conventional strength training sets? (stimulating reps or high effort reps?)
Muscular failure is reached when the combination of the available fatigue mechanisms causes insufficient whole muscle force to be produced by the muscle and the CNS such that the weight can be lifted. All muscular contractions involve peripheral fatigue mechanisms as well as both kinds of CNS fatigues, (supraspinal and spinal). Nevertheless, it is possible to alter the relative contributions of each type of fatigue mechanism, and this means that stimulating reps are not equal to high effort reps (even in the context of self-selected tempo strength training).
Stimulating reps are those that involve maximal levels of motor unit recruitment (which ensures a very large number of muscle fibers are active) while the working muscle fibers simultaneously produce a high level of mechanical tension by shortening relatively slowly (due to the force-velocity relationship). This combination allows a very large number of muscle fibers to experience high levels of mechanical tension, and therefore grow.
High effort reps are those in which a high perceived effort is achieved, and this category of reps is much larger than the category of stimulating reps, insofar as it is possible to exert a high effort during a rep without that rep stimulating any hypertrophy. This can occur either  because of the load on the bar, and therefore the muscle fiber shortening velocity, or  because of the presence of an excessive amount of CNS fatigue.
Indeed, high effort reps can be performed in a unfatigued state or in a fatigued state, and with a range of loads. In an unfatigued state, high effort reps with heavy loads are the same as stimulating reps, but high effort reps with light loads are not. High effort reps in an unfatigued state with light loads (as occurs during vertical jumping or throwing) involve maximal levels of motor unit recruitment but do not cause muscle growth, because the mechanical tension experienced by the working muscle fibers is too small. In a fatigued state, high effort reps are usually stimulating with both heavy and light loads, but when fatigue reaches a very high level (such as occurs when performing multiple sets with short rests or towards the end of a workout), then the presence of CNS fatigues (and possibly also calcium ion-related fatigues) can impair the stimulus for hypertrophy.
Essentially, there are stimulating reps that involve very high levels of motor unit recruitment at the same time as high levels of single muscle fiber mechanical tension, and there are non-stimulating but still high effort reps that may involve either inadequate motor unit recruitment levels (because of the presence of CNS fatigues) or inadequate mechanical tension (owing to a fast muscle fiber shortening velocity).
Having established this important point, we can now address the effects of intra-set rest periods on hypertrophy.
How does fatigue differ between cluster sets and conventional strength training sets? (cluster sets, finally)
Importantly, unless we differentiate between  high effort reps with a slow bar speed that are non-stimulating because of the negative effects of fatigue mechanisms, and  stimulating reps, there is nothing within the existing stimulating reps model that enables us to differentiate between main sets during conventional strength training and subsets performed within 5 or 6 reps of failure during cluster set strength training.
For example, if a cluster set training program involved 3 sets of 8 reps with a 5RM load, subdivided into four clusters of 2 reps per set with 30 seconds of rest between clusters (or rest as needed to allow the reps to be performed), then this should theoretically cause 8 stimulating reps. A corresponding conventional strength training program of 3 sets of 8RM would probably only cause five or six stimulating reps. This would suggest that there is great benefit in using cluster sets for bodybuilding. The same analysis can be performed with drop sets and rest pause training, which both involve a very similar approach. A typical drop set or rest pause training program might involve 1 set of 5 reps to failure with 3 subsets per set (which is 15 reps per set) and the unadjusted stimulating rep model would predict that this would achieve the same results as 3 sets of 5 reps to failure during a conventional strength training program. This clearly does not happen, as a couple of recent studies have revealed.
It seems likely that there is only marginal benefit of the extra subsets during drop set and rest pause training programs, in comparison with a single set of conventional strength training. The reduced impact of the later subsets during drop set training, rest pause training, and likely also cluster set training must arise due to the negative effects of too much supraspinal or spinal CNS fatigue or too much excitation-contraction coupling failure building up over the course of the exercise bout, in much the same way as these fatigue types build up when strength training with short rest periods, thereby reducing the amount of hypertrophy that is stimulated.
Nevertheless, there is an important practical difference between cluster set training and both drop set and rest pause training configurations. During drop set and rest pause training, each subset is performed to muscular failure, whereas most of the subsets are not typically performed to muscular failure during cluster set training configurations. This is important because both supraspinal CNS fatigue and excitation-contraction coupling failure are only really provoked as muscular failure approaches. These are two key fatigue mechanisms that suppress the ability of the muscle fibers within the muscle to stimulate hypertrophy.
Therefore, while drop set training does not appear to be an efficient or particularly effective method for increasing the number of stimulating reps that are performed in any given period of time, cluster sets may yet provide an opportunity to increase the number of stimulating reps per set, to the extent that supraspinal CNS fatigue and excitation-contraction coupling failure can be avoided. Reducing supraspinal CNS fatigue means avoiding metabolite accumulation within the muscle in order to reduce afferent feedback, which means lifting either heavy (1–5RM) or moderately-heavy (6–10RM) loads and avoiding training to failure in the early subsets of a cluster set. Reducing excitation-contraction coupling failure similarly means avoiding training within two reps of failure in the early subsets of a cluster set.
For example, a cluster set might involve 3 subsets of 2 reps with 5RM with a short rest to allow muscular failure to be reached on the final rep of the third subset. This might enable six stimulating reps rather than the five that would be achieved with a standard strength training configuration of a single set with 5RM. With a longer rest between the subsets, 4 subsets might even be possible, depending on the recoverability of the lifter from short-term fatigue mechanisms, thereby allowing eight stimulating reps.
What does this mean in practice?
Although very little research has been done into cluster sets where the final cluster is taken to muscular failure, an analysis of the fatigue mechanisms suggests that there could be a small, theoretical benefit of this way of training for achieving hypertrophy. Cluster sets could be used to increase the number of stimulating reps per set, and thereby increase the efficiency of strength training sets for muscle growth. Nevertheless, this benefit is likely only to be achievable if early clusters are not taken to failure and the loads are either heavy (1–5RM) or moderately-heavy (6–10RM) such that the main, negative fatigue mechanisms of supraspinal CNS fatigue and excitation-contraction coupling failure do not interfere as much as they do during drop set training and rest pause training methods.
If you enjoyed this article, you will like my second book (see on Amazon).