How to count training volume and design a sensible training split

People talk a lot about how much training volume to do and which exercises to do but rarely about their relation: how do we count the volume of different exercises when it comes to stimulating certain muscle groups? Do you count bench presses towards your triceps volume, for example? Should you count deadlifts for your back or legs? Do horizontal pulls count for the biceps? In this article I’ll explain why only 1 of those 3 questions actually makes sense and how you should calculate your training volume for different exercises based on the current science. Since the allocation of volume is strongly intertwined with the type of training split you do, we also have to discuss how to design a logical training split.

Method 1: Horizontal & vertical pushing & pulling

A popular method to segment volume is to divide exercises into horizontal or vertical pushing or pulling movements and then equating the volume of pushing and pulling exercises in the horizontal and vertical planes. So for every set of a horizontal pushing exercise, you should do a set of a horizontal pull. This method can be viable, but it’s an extremely simplified take on human functional anatomy. Not all horizontal/vertical pushing/pulling exercises are functionally interchangeable. Below are 3 examples of how exercises in the same category can stimulate different muscles. Likewise, exercises in different categories can stimulate the same muscles.

1. Dips and overhead presses are both vertical pushing exercises, but they train completely different muscles. Same category, different training stimulus. Dips are functionally more related to most horizontal than vertical presses.
2. Upright rows and wide grip pull-ups are both vertical pulling exercises, but again they train completely different muscle groups. Same category, different training stimulus. Upright rows are functionally more related to most vertical presses than pulls.
3. Standard barbell shrugs and overhead shrugs train very similar musculature, in particular the upper traps, but the standard version is a pull, whereas the overhead version is a press. Different categories, similar training stimulus.

For the lower body, things get even more confusing. Are skater squats a press or a pull? They’re technically a pull if you hold a weight in your hands, yet they become a press if you put a weight on your back, yet this doesn’t change the training stimulus much. Hip thrusts are a vertical… something, but they’re very different from both standard vertical pulls like deadlifts and vertical presses like squats. And what about leg presses? They’re obviously a press, but they can be both vertical or horizontal, depending on the machine.

And what are rotary movements like hyperextensions, leg extensions and lateral raises? As the weight is not coming closer to you nor is it moving away from you, they are neither. Most people solve these issues by figuring out which muscles are trained and then grouping those together, which completely defeats the purpose of the whole classification system. If you have enough understanding about functional anatomy to do this, I recommend you don’t bother with the whole horizontal/vertical push/pull categorization to begin with. Continuing to use the system despite knowing it makes little sense is like a company continuing to use outdated software: everyone at the office has to bend over backwards to accommodate the flaws of the system while eventually the upgrade is coming anyway.

If you’ve ever worked in an office, you need to watch Office Space.

And if we’re going to classify exercises by with muscles they train rather than their movement pattern, let’s look at bodybuilding splits.

Method 2: The bodybuilding body part split

Rather than have to figure out whether lateral raises and deadlifts should go on your pull or push day, you could do what bodybuilders have traditionally done. Group all exercises for the same muscle groups together and train those on 1 day per week. Most research finds more advanced trainees most likely need to train a muscle at least twice per week, and possibly more, for maximum muscle growth, but let’s leave that aside for now. With 1 day for each body part, volume should be similar for all muscles… right?

It could be, if the division of body parts was strict and made anatomical sense. Neither is generally the case. First, the ‘back’ is not a muscle, m’kay? You should distinguish between at least the lats, the erector spinae, the rear delts and the traps. Even worse is classifying ‘legs’ as 1 muscle group. You should distinguish between at least the glutes, the quads, the hams and the calves. For example, squats don’t train ‘legs’, as they don’t train the hamstrings appreciably. Hamstring activity during squatting movements is low, so the hamstrings grow barely 20% as fast as the quads and the glutes after squatting [2, 3]. Plus, the lower body makes up about half your body mass, so if you do 1 day for that half but you have a separate day for shoulders or arms, which are comparatively tiny muscles, don’t be surprised if you end up in one of those memes of bros with chicken legs. And ladies, if you primarily care about lower body development, you almost certainly want to train your legs/hips at least twice per week. A single leg day per week isn’t going to cut it. I think leg days or lower body days are often not ideal anyway. They take forever and they’re disproportionately brutal compared to upper body days, especially if you do squats and deadlifts on the same day.

The second major problem with bodybuilding body part splits is the lack of distinction between body parts. Many bros train their arms and shoulders way more than other muscles, because in addition to their isolation work or even ‘arms day’, they already get trained by many compound exercises. Just because your primary goal of doing the bench press is to build Schwarzeneggerian pecs, doesn’t mean the front delts and the triceps don’t experience mechanical tension. These problems have resulted in bodybuilding body part splits to become known as ‘bro splits’.

Method 3: Sets per week per muscle group

The logical improvement over the bodybuilding split is to count the total number of effective sets for each muscle group per week (or whatever unit of time you prefer). The total number of work sets per week per muscle group is the most effective measure of volume for muscle growth we have. This may sound self-evident to many of my long-term readers now, but for decades, lifters didn’t fully realize that muscle growth is an almost purely local process and didn’t have a systematic approach to count volume. Many lifters had this idea that a squat, for example, was this magical mass builder that jacked up your anabolic hormone levels to the stratosphere and fueled whole-body gains. Nope. It’s just a good exercise to stimulate mechanical tension in the quads, glutes and erector spinae. In fact, to my knowledge, I was one of the first in the fitness industry to popularize the idea over a decade ago that training program design should be muscle-specific. Effectively, you should create a training program for each individual muscle and then combine those together into a training split and make sure all training volumes are optimal at the same time.

To reach a certain number of stimulating sets for a muscle, you have to assign training volumes to exercises to denote which muscles are effectively trained by 1 work set of that exercise. (You can generally ignore warm-up sets.) For example, a set of Bayesian curls should arguably count as 1 set for the biceps and nothing else. Compound exercises are trickier, because if you want to be precise, you’ll have to assign fractional volumes to exercises. For example, dumbbell rows stimulate only approximately half the muscle growth in the biceps as dumbbell biceps curls do. Due to the limited demand for elbow flexion torque, biceps activity during rows isn’t very high and we see limited muscle damage and neuromuscular fatigue after dumbbell rows compared to dumbbell curls (although contested here). Thus, 6 sets of dumbbell rows should only count as 3 sets of effective biceps work. And you probably won’t maximize biceps growth with just rows, as you won’t activate the highest threshold motor units much.

Similarly, you should probably count barbell bench presses and push-ups with a regular grip as 50% for the triceps, because these presses stimulate approximately twice as much muscle growth in the pecs as in the triceps on average. You can see the relevant studies in a research overview from my PT Course here.

To be even more precise, you should count barbell bench presses as 100% for the pecs and the lateral and medial heads of the triceps but barely for the long head. The long head is a bi-articulate muscle. It doesn’t just extend the elbow. It also extends the shoulder, bringing the elbow down. Since you want to raise the elbow during a bench press, contracting the long head produces a mixture of desirable and undesirable forces. Therefore, the body probably doesn’t fully activate the long head of the triceps during bench presses, although not all research finds significant differences in muscle activity between the heads. (EMG research is noisy.) Unbeknownst to many lifters, lat prayers (a form of pull-overs) activate the long head of the triceps better than bench presses. As a result, barbell bench presses stimulate significantly less muscle growth in the long head of the triceps than skull-crushers: 2% vs. 18%. Lat prayers and barbell bench presses thus combine very well: if you do 3 sets of each, you can count that as 3 sets of complete triceps work.

Regional hypertrophy is also very evident for many other muscle groups. For example, you should separate the delts into at least the front, middle and rear parts and the traps into the upper, middle and lower parts. These parts have clearly different, sometimes even opposing, functions.

Exercise technique also plays an important role in determining which muscles are effectively trained by an exercise. For example, a very wide grip bench press reduces triceps activity, whereas a very narrow grip with your elbows tucked into your sides reduces lower pec activity. In contrast to popular belief, a narrow grip bench press is still a poor triceps exercise though. Your elbow position also significantly influences which heads of the delts you train. The way most people perform their overhead presses is very front-delt dominant.

However, you don’t have to overcomplicate things. Don’t bother with crazy fractionals like counting 84.9% of the volume of an exercise. The science isn’t there yet to be that precise. If a muscle is causing a particular movement (so it’s an agonist or a synergist), and you don’t have a good reason to doubt that the muscle is subject to high tension, you can probably count it as being 100% stimulated. For example, many people think compound exercises are inherently worse than isolation exercises because they’re an indirect exercise or whatever broscience jargon is used, but this is not true. A 2023 meta-analysis by Rosa et al. found that single-joint and multi-joint exercises, which generally meant isolation vs. compound exercises, resulted in similar muscle growth on average in the literature. For example, Gentil et al. (2015) compared pronated grip, wide lat pulldowns to supinated grip barbell curls and found that they resulted in equal biceps growth and strength gains (isokinetic peak elbow flexor torque). While the study didn’t have major statistical power with only 10 subjects training twice per week for 10 weeks, these results suggest any exercise that stimulates enough mechanical tension on the muscle fibers can achieve major muscle growth. Correspondingly, Pompermayer et al. (2021) found that pulldowns and biceps curls stimulate a similar level of muscle damage in the biceps. Pulldowns involve elbow flexion, so the brain will recruit the elbow flexors to perform the movement, whether you feel them or not.

Moreover, you can probably round down anything below 40% stimulus to 0%. Muscles require a minimum tension that seems to be equivalent to around 40-60% of maximum voluntary isometric contraction strength to get stronger.

Conclusion

In conclusion, muscle growth is best predicted by the number of effective sets you do for that muscle group per week. How effective an exercise is for a particular muscle group, or part of the muscle, depends on the exercise’s biomechanics and the exercise technique you use. There’s no one-size-fits-all answer. To optimize a training program, you should assign a percentage to each exercise that denotes how well it stimulates each involved muscle. Then you need to combine exercises with the numbers of sets that results in your target number of total sets for each muscle group. How you call the resulting training split doesn’t matter. Don’t let yourself get limited by simplistic categories like push/pull/legs. Whatever program optimizes your training volume, frequency and other program parameters is ideal for you. Optimal program design requires some puzzling, but the reward is maximum muscle growth throughout the body and an aesthetically balanced physique.