If you’ve been lifting weights for a while, you’ve probably heard this at least once: “Get bigger, get stronger.” It sounds simple enough. But if you dig into the research on how muscle growth relates to strength gains, things get messy fast.
Here’s the short answer: hypertrophy does contribute more to your strength than older studies suggested — but it is not anywhere near as dominant as some recent headlines claim. The relationship depends heavily on who you are and how you measure it.
What We Thought We Knew
Until recently, the research told a pretty consistent story. In people new to training, muscle growth explained very little of the variance in strength gains. But as lifters got more experienced, that link grew stronger and stronger.
In studies on untrained subjects, we typically see R² values near zero — meaning hypertrophy barely explains the differences in how much strength each person gains. Give those same people a few months of training first, and R² climbs into the 0.2–0.3 range. In well-trained lifters, it can reach around 0.7, so hypertrophy accounts for roughly 70% of the variance.
Here’s what I tell my clients: this pattern makes intuitive sense. Early strength gains come mostly from getting better at the movements — improved technique, motor learning, neural drive. In studies on untrained lifters, the average muscle size increase is about 5% while strength jumps roughly 22%. Even if hypertrophy plays a causal role, it can account for less than a quarter of those early gains.
As training experience increases, there are fewer technical improvements left to capture. So hypertrophy explains more and more of what happens next.
The Study That Seemed to Change Everything
A study by Marques and colleagues appeared to flip this understanding on its head. The paper is titled Muscle Growth Is Very Strongly Correlated with Strength Gains after Lower Body Resistance Training: New Insight from Within-Participant Associations.
The researchers argued that previous work underestimated the hypertrophy-strength link because it relied on between-subject analyses — basically, comparing one data point per person (the change from pre-training to post-training). They proposed that a repeated-measures correlation, which looks at within-participant changes over time, is a better tool for this job.
Think of it this way. Imagine two people with biceps of the same size. Person A has favorable tendon insertions and high specific tension — their muscle fibers generate more force per unit of cross-sectional area. Person B has the opposite. If both gain 5% in biceps size, person A might gain 20% in strength while person B gains only 10%. Plot those two data points side by side, and it looks like a weak relationship. But within each individual, the same muscle growth clearly drove their own strength increase.
Prior work supported this idea. A study by Vigotsky and colleagues found that between-subject correlations showed muscle size explaining less than 5% of strength variance in untrained subjects. But a hierarchical linear model — which allows each person to have their own baseline and trajectory — explained 7.4–24.1% of the variance (an r-value of roughly 0.25–0.50). Within-subject methods consistently outperformed between-subject ones.
What the Marques Study Actually Found
Thirty-nine untrained men completed 15 weeks of resistance training focused on the quadriceps. Strength was measured as knee extension 1RM and maximum isometric torque; muscle size as quadriceps volume via MRI.
On average, isometric knee extension strength increased by 21.6%, knee extension 1RM rose by 28.6%, and quadriceps volume grew by 12.7%.
The repeated-measures correlation between hypertrophy and strength came in at r = 0.92 for isometric torque and r = 0.89 for 1RM. That translates to roughly 80–85% of the variance explained.
By contrast, the traditional between-subjects correlations sat in the range of r = 0.35–0.60 — meaning only about 12–35% of variance explained by hypertrophy using older methods.
At first glance, this is a striking result. It would suggest that previous researchers missed an almost perfect relationship between muscle growth and strength in beginners simply because they used the wrong statistical lens.
The Catch: Repeated-Measures Correlations Run Hot
So did prior research really underestimate the link by this much?
Not exactly. The first red flag appears when you compare Marques to Vigotsky. If the problem with earlier studies was just the use of between-subject methods, then Vigotsky — which also used a within-subject approach — should have found similarly sky-high correlations. In fact, Vigotsky’s hierarchical linear model allowed even more flexibility than Marques’ repeated-measures method, since it permitted both intercepts and slopes to vary between subjects.
Yet the Vigotsky study found that within-subject analyses explained less than 25% of the variance in strength gains. That is a meaningful gap.
I Ran the Numbers Myself
To figure out what was going on, I asked a simple question: how strong would the repeated-measures correlation appear if hypertrophy and strength were actually unrelated?
You might assume the answer is r = 0. But it is not.
I randomly generated 5,000 simulated “subjects” matching the Marques study’s summary statistics — same means, standard deviations, change scores, and change score SDs for knee extension 1RM, isometric torque, and quadriceps volume. By design, changes in muscle size were completely independent of changes in strength. This simulation lives in a universe where hypertrophy has zero impact on strength gains.
Even with absolutely no relationship between the two variables, the repeated-measures correlation returned r-values of 0.81–0.83. That means it would look like hypertrophy explained roughly 65–70% of the variance — even when hypertrophy had nothing to do with strength at all.
The reported r-values of 0.89 and 0.92 suddenly look less impressive. They are higher than the null case, yes. But they imply that hypertrophy explains only about 12–17% more variance (additively) beyond what random noise would produce.
To put it another way: if you treat an R² of roughly 0.67 as your effective zero, then the Marques results tell us hypertrophy explains about 37% more of the remaining unexplained variance for 1RM strength, and about 53% more for isometric torque.
Here’s what I think the fairest read of these numbers looks like: hypertrophy probably explains around 20–25% more of the variance in strength gains than between-subjects correlations on change scores would suggest. That is meaningful — it means older methods do undercount the link. But repeated-measures correlation, when interpreted without caution, overshoots to a hilarious degree.
To hammer this home: the Marques study’s r = 0.89 for hypertrophy and 1RM does not differ significantly from the null case of r = 0.82 (the confidence interval runs from 0.81 to 0.94, so p > 0.05). We simply cannot be confident that 0.89 represents an association meaningfully stronger than what we’d expect purely by chance.
Why This Matters for You
The research says hypertrophy matters more for beginners than we once thought. But here’s the practical part: this does not change how you should train.
Here’s what I tell my clients when they ask whether they should focus on building size or chasing strength numbers: you do both at the same time, because training that stimulates growth also trains your nervous system to handle more load. The debate over exact correlation coefficients is interesting — it matters for how scientists interpret data — but it does not mean you need to pick one goal over the other.
The Marques study ran longer than most beginner studies (15 weeks), which produced notably more hypertrophy and slightly larger strength gains than usual. One of their strength tests required minimal skill — maximal isometric knee extension torque — reducing noise from learning effects. Those design choices helped, but they also mean the results may not translate directly to every exercise or every training program you follow.
I also want to note that even the “old school” between-subjects correlations in this study (r = 0.35 for 1RM, r = 0.60 for isometric torque) were stronger than we typically see in untrained lifters. Longer study duration and larger actual gains naturally produce cleaner signals.
A Word of Caution on Repeated-Measures Correlations
I strongly suspect repeated-measures correlations will start showing up more often in exercise science publications. When you see one, do not just read the r-value the way you would a standard Pearson correlation.
Strength and hypertrophy outcomes tend to have coefficients of variation around 1.0 — meaning the standard deviation of changes is roughly equal to the mean change. When both variables are positive and their CVs fall between 0.5 and 1.5, repeated-measures correlations in the range of 0.65–0.85 may actually signal little to no association at all.
To illustrate: I created a quick simulation where two dummy variables increased completely independently — one by 1 ± 1 unit, the other by 3 ± 1 units. No relationship whatsoever. And yet the repeated-measures correlation came in at r = 0.83.
The evidence here is genuinely nuanced. Repeated-measures correlations are a useful tool, but they require more careful interpretation than most readers — and sometimes researchers — give them credit for.
The Science, Simplified
When does muscle growth predict strength gains? In trained lifters, hypertrophy explains up to about 70% of the variance in how much strength different people gain. In beginners, older methods suggest it explains very little — maybe 12–35%. A new repeated-measures method pushes that number to 80–85%, but simulations show those values are inflated: roughly half of that correlation exists even when hypertrophy and strength have no real relationship.
What should you do? Train for both size and strength together. The exact correlation coefficient does not change the practical recommendation: progressive overload with adequate volume builds muscle and gets you stronger at the same time. If you want to understand research findings more deeply, pay attention to how the data was analyzed — a high r-value from repeated-measures methods deserves extra scrutiny.
The numbers will keep getting refined. Your next workout is what will move the needle. Pick a program that challenges both your muscles and your strength, run it for six to eight weeks, and judge the results yourself. That has always been the secret, and it remains so.
Leave a Reply