TriPeak

Polarised training

Most of it easy, some of it very hard, almost none in between.

How the week is distributed

A polarised week puts roughly 80% of training time easy, 5% moderate and 15% hard. The percentages are of time, not of sessions — which is why a week with two hard sessions in it can still be 80% easy.

Easy

80% of weekly training time

Moderate

5%

Hard

15%

How it plans

• Keep roughly four fifths of weekly training time genuinely easy — below the first ventilatory threshold, conversational, slower than feels satisfying.
• Make the hard sessions properly hard: VO2-max intervals and above, two or at most three a week.
• Avoid the middle. Tempo and threshold work is the first thing cut, not the last.
• Easy days are the training, not the filler between it. Do not let them drift up.
• Volume rises before intensity does.

Why it is meant to work

The argument starts from an observation: when the training of elite endurance athletes is measured rather than recalled, very little of it sits at threshold. Most is easy enough to hold a conversation through, and what is left is genuinely hard. The middle, where it is tempting to spend every session, is largely absent.

The physiological case for why that might be right is that the two ends do different jobs. Long easy work drives the peripheral adaptations — capillary density, mitochondrial volume, fat oxidation, stroke volume — at a cost in fatigue low enough to be repeated daily. Hard intervals drive the central ones, chiefly maximal oxygen uptake, and need to be genuinely hard to do so. Threshold work is argued to sit awkwardly between: too taxing to do daily, not intense enough to maximally stress the top end.

The practical claim is therefore about recovery as much as stimulus. Keeping the easy days truly easy is what makes two or three genuinely hard sessions a week survivable, and most amateurs fail this not by training too little but by making their easy days moderate and their hard days moderate too.

Who it suits, and how it fails

Suits

Athletes with the hours to make the easy volume count, and a training age high enough that hard days are genuinely hard.

Watch out for

Needs volume to work. On six hours a week, eighty percent easy is not much stimulus, and this is the philosophy most often applied by people who do not have the hours it assumes.

Strength of evidence

Rests largely on observational work describing what elite endurance athletes actually do, plus a handful of small trials favouring it over threshold-focused training. Descriptive rather than prescriptive: that elites train this way does not establish that it made them elite.

The research behind it

Every citation below was checked rather than remembered, and each note says what the study found and what it does not establish.

1. Seiler & Kjerland (2006), Scandinavian Journal of Medicine & Science in Sports 16(1):49–56The paper the whole idea is built on. Eleven junior cross-country skiers, 384 sessions over 32 days, measured by heart rate and session RPE. Found they trained remarkably little at threshold. It is descriptive and small: it says what these athletes did, not that doing it made them fast.
2. Stöggl & Sperlich (2014), Frontiers in Physiology 5:33The most cited trial in favour. 48 well-trained endurance athletes randomised across four models for nine weeks; the polarised group improved most on VO2peak and time to exhaustion. Nine weeks is short, the groups were small, and its design has been argued over since.
3. Rosenblat, Perrotta & Vicenzino (2019), Journal of Strength and Conditioning Research 33(12):3491–3500Meta-analysis of polarised against threshold. Found a moderate effect favouring polarised for time-trial performance — from only four studies, and rated the evidence base as limited. This is the honest summary of where things stood.
4. Comparison of Polarized Versus Other Endurance Training Intensity Distributions (2024), Sports MedicineA more recent and broader meta-analysis. Worth reading precisely because it tempers the earlier enthusiasm: the differences between distributions are smaller than the arguments about them.