Polarized versus pyramidal: where the advantage exists and where it does not

The argument about intensity distribution is one of the longest-running in endurance training. Some point to the elite, who run slowly almost all the time and very fast occasionally. Others point out that middle-zone work has not gone anywhere and works excellently for many people. A recent meta-analysis finally lets us talk in numbers rather than slogans. The answer, though, comes with three caveats.

Three models, briefly

All the schemes describe the same thing: how the hours are distributed, not the sensations. The three-zone model divides work into low (below the first threshold), moderate (between the thresholds) and high (above the second threshold) intensity.

Polarized (POL): 75–80% of the time in zone one, 15–20% in zone three and less than 10% in zone two. Lots of easy, some very hard, almost nothing in between.
Threshold (THR): more than 35% of the time in zone two — that is, an emphasis specifically on “moderate” work.
Pyramidal (PYR): all three zones are used, volume decreases from low to high intensity, but the middle zone takes up a noticeable share — more than in polarized and less than in threshold.

The defining feature of polarization is the deliberate rejection of the middle zone. That is exactly what was tested.

What the meta-analysis showed

The systematic review with meta-analysis gathered 17 controlled studies and 437 participants. Intervention duration ranged from 4 to 16 weeks, with a median of 10.

For VO₂peak, polarization really does win: SMD = 0.24 (95% CI 0.01–0.48; p = 0.040). The effect is small, but on the GRADE scale the authors rated the certainty of the evidence as high — a rare situation in training research.

And then comes the most interesting part: the subgroups.

Caveat one: only short blocks. The advantage held only in interventions lasting less than 12 weeks (SMD = 0.40; 95% CI 0.08–0.71; p = 0.01). In blocks of 12 weeks and longer the difference vanished entirely (p = 0.83).

Caveat two: only in trained athletes. In highly trained, national-level athletes the gain was noticeable (SMD = 0.46; p = 0.01). In merely trained and developing athletes there was no difference at all (p = 0.82).

Caveat three, and it is the main one. For every measure other than VO₂peak, no differences were found whatsoever:

time trial — SMD = −0.01 (95% CI −0.28 to 0.25; p = 0.92);
time to exhaustion — SMD = 0.30 (95% CI −0.20 to 0.79; p = 0.24);
speed or power at the second threshold — SMD = 0.04 (95% CI −0.21 to 0.29; p = 0.75).

In other words: polarization raises a laboratory measure slightly faster, but in the thing we actually train for — the result over a distance — no advantage is visible.

How a runner should read this

The practical translation turns out unexpectedly calm.

There is no universal winner. If a pyramidal scheme works for you, there is no reason to change it in favour of polarization: no difference was shown in time trials.
Polarization is a short-block tool. The subgroup logic suggests how to use it: 6–10 weeks of a polarized block as a way to nudge aerobic power, rather than as a permanent way of life.
The higher your level, the more it makes sense. In beginners and mid-pack athletes almost everything improves from almost anything, and the structure of the distribution is far from the most important lever. Consistency and volume come first.
The sequence may be more interesting than the choice. A separate 16-week study in well-trained runners compared the pyramidal and polarized models and found that the biggest improvements in VO₂peak, thresholds and 5 km performance came from switching between schemes: a pyramidal block first, then a polarized one. That is closer to classic periodization than to picking one “correct” model forever.
Count by time, not by number of sessions. One threshold session a week lasting an hour is already a noticeable share of the volume for an amateur training six hours a week. Many plans that look “polarized” on paper turn out to be pyramidal in practice.

Limitations

The authors list them in detail, and they are substantial. Some of the included studies did not report the exact distribution percentages at all, which complicated classification. Groups could differ not only in intensity but in other important variables that were not accounted for. Time trial distances ranged from 100 metres to 40 kilometres — hard to compare within a single analysis. And finally, questions that decide a great deal in real life were left out of scope: how the different models affect recovery and the risk of overtraining.

Key points

Polarized model: 75–80% in zone one, 15–20% in zone three, less than 10% in zone two. The pyramidal model gives the middle zone more room.
For VO₂peak polarization wins: SMD 0.24, with high certainty of evidence.
But only in blocks shorter than 12 weeks (in long ones there is no difference) and only in highly trained athletes (in everyone else, none).
In time trials, time to exhaustion and at the second threshold there is no difference at all — meaning the advantage did not show up in results.
The sensible application is a short polarized block within periodization, not a lifelong model.
Count the distribution by hours, not by number of sessions: for an amateur, one long threshold session easily breaks the “polarized” nature of a plan.

Sources: “Comparison of Polarized Versus Other Types of Endurance Training Intensity Distribution on Athletes' Endurance Performance: A Systematic Review with Meta-analysis”, Sports Medicine, 2024. https://doi.org/10.1007/s40279-024-02034-z; https://pmc.ncbi.nlm.nih.gov/articles/PMC11329428/. “Effects of 16 weeks of pyramidal and polarized training intensity distributions in well-trained endurance runners”. https://pmc.ncbi.nlm.nih.gov/articles/PMC9299127/