Polarised Training Method Explained

Your weekly hours9h

Fits your week. Coherent from 8 to 25 hours a week.

Where the time goes

Compare with

By time across all three sports. The threshold bar is the swim set at CSS, reported outside the polarisation ledger. Seiler counts 80/20 by session.

A representative week

Show the week for

Across all three sports.

Mon

Swim 40 min1,800–2,000 m, technique-led, nothing above CSS. Take it as a full rest day if the previous week was heavy — Seiler's own template is Monday rest or 60 min easy.

Tue

Bike 75 min — 5 × 4 min20 min easy, then 5 × 4 min at 268–285 W (107–114% FTP), each rep followed by 4 min easy spinning, then 15 min cool-down. Build 10–25 W across the set and finish with a rep in reserve. Hard session 1 of 2.

Wed

Swim 30 minEasy aerobic. Swim first on any shared day so form is not compromised by fatigue.

Run 50 minEvery kilometre conversational, roughly 5:20–5:40/km against a 4:20 threshold. Walk the hill if heart rate pushes past the VT1 cap. This is the day the method lives or dies on.

Thu

Swim 55 min with a threshold main set400 m easy, 8 × 50 m drill off 20 s, then 8 × 200 m at about 1:40/100 m (CSS) off 20 s rest, 400 m easy — about 2,800 m, of which 27 min is at threshold. The swim will not polarise and is not made to: this spends whole-athlete quality budget but is reported outside the polarisation ledger, and it is the threshold bar on the chart.

Bike 45 minEasy spin, well below VT1, as far from the swim as the day allows.

Fri

Run 60 min — 7 × 400 m20 min easy, then 7 × 400 m at 3–5 km effort (about 75–80 s) with 75 s jog after each — about 18 min — then 22 min easy. The canonical Norwegian model is 10–15 reps; seven here because interval work is capped at 8% of weekly run volume and this athlete runs about 36 km. Hard session 2 of 2.

Sat

Bike 2 h 45 minContinuous below VT1 at 60–72% HRmax. Add six to eight maximal 5-second sprints at least 20 min apart. Ride alone or ride at your pace — a group ride turns this into tempo and the week stops being polarised.

Sun

Run 85 minLong and easy, same cap as Wednesday. On three runs a week it is a bigger share of the running than Daniels' 25–30% guide, which assumes six or seven — so it is bounded by duration and by staying under VT1, not trimmed. Expect heart rate to drift up in the last half hour as the VT1 boundary itself falls; hold the pace down rather than chasing the number.

What Tuesday looks like

Bike 75 min. 20 min easy, then 5 × 4 min at 268–285 W (107–114% FTP), each rep followed by 4 min genuinely easy spinning, then 15 min cool-down. Build across the set — first rep at the bottom of the band, last 10–25 W higher — and finish able to hold the target for one more. That is 20 min of accumulated work, the entry dose. Progress toward Seiler's 30–40 min by lengthening the reps and dropping the watts with them — 5 × 6 min at 263–275 W (105–110% FTP), then 4 × 8 min at 250–263 W (100–105% FTP) — rather than by adding a third hard day. A four-minute band on an eight-minute rep is a session nobody finishes. Then Wednesday is actually easy: 50 min running with every kilometre conversational, around 5:20–5:40/km against a 4:20 threshold. Walk the hill if heart rate pushes past the VT1 cap. Wednesday is where this method is won or lost.

Signature sessions

4 × 8 min at ~90% HRpeak, easy spinning between (about 2 min — the published paper does not state the recovery duration). The best-evidenced session in the family: Seiler, Jøranson, Olesen & Hetlelid (2013) randomised 35 trained cyclists averaging ~6 h/week to seven weeks of twice-weekly 4 × 4, 4 × 8 or 4 × 16 at maximal tolerable intensity. Achieved intensities were 94±2, 90±2 and 88±2 %HRpeak with 13.2, 9.6 and 4.9 mmol/L lactate. 4 × 8 gained 11.4% on a composite of VO2peak, power at VO2peak and power at 4 mM, against 5.5% (4 × 4), 5.6% (4 × 16) and 4.2% for low-intensity only. Their conclusion: accumulating 32 min at 90% HRmax beats 16 min at ~95% despite lower RPE. Prescribe it as the study ran it, maximal tolerable intensity held across the whole set. If a wattage is wanted, eight-minute reps sit at 250–263 W (100–105% FTP) for the reference athlete — derived from FTP, not from the paper.

4 × 4 min at 90–95% HRpeak with 3 min active recovery, inside a 20 min warm-up and 15 min cool-down both at 75% HRpeak. The exact protocol run by the polarised group in Stöggl & Sperlich (2014).

Norwegian bike pyramid: 2, 4, 6, 8, 10, 8, 6, 4, 2 min with 1–3 min easy between (Tønnessen et al. 2024, listed as a Zone 4 session).

Running above VT2, from Tønnessen et al. 2024. Zone 4 (88–92% HRmax, 4.0–6.0 mmol/L, 15–35 min accumulated work): 4–6 × 1000 m at 2:30–3:00 with 2 min recovery; 10–15 × 400 m at 60–70 s with 60–75 s; or 4–6 × 2 km at 6:00–6:20 with 2 min. Zone 5 (>93% HRmax, 6.0–10.0 mmol/L, 10–20 min accumulated): 5 × 300 m in 35–40 s with 3–5 min, or 10–12 × 200 m in 25–30 s with 30–60 s. Those paces are elite — keep the rep count and recovery structure, run the reps at the athlete's own 3–5 km effort, and cut rep count to stay inside the run dose caps.

Never write 400s off 30–45 s, or 1000s off 90 s, and call them polarised. Short recoveries on short reps are Norwegian Zone 3 threshold work at 2.5–4.0 mmol/L — which is Seiler Zone 2, the exact zone this method exists to minimise. Prescribing that as a VO2max session puts the athlete in the middle while telling them they are polarised.

Long easy session: 150–240 min continuous below VT1 at 60–72% HRmax. Stöggl's polarised group added six to eight maximal 5-second sprints separated by at least 20 min — cheap neuromuscular work that does not disturb the aerobic session.

Swim: do not force the pool to polarise. Tønnessen's published session models (competitive swimmers, not triathletes) run Zone 3 as 4–6 × 1000 m in 11:00–12:00 off 1:30–2:00, or 20–24 × 200 m in 2:05–2:20 off ~30 s; Zone 4 as 4 × 800 m in 8:30–9:30 off 2:30, or 6–8 × 400 m in 4:10–4:40 off 60 s. Rescale to the athlete's CSS — at 1:40/100 m, a Zone 3 200 sits near 3:20 — and keep the swim out of the polarisation ledger.

Who it suits

Athletes with 8+ h/week in a single discipline, or 10+ h/week of triathlon with real per-sport volume, who have the time to accumulate easy hours and the discipline to actually go easy on the easy days. Competitive and well-trained athletes specifically — Rosenblat's subgroup finding is that competitive athletes may gain more VO2max from polarised while recreational athletes gain more from pyramidal, and Silva Oliveira's meta-analysis found the small VO2peak advantage only in highly trained/national-level athletes (SMD 0.46, p=0.01) and only in interventions under 12 weeks (SMD 0.40, p=0.01). Events raced at or above threshold suit it best: 1500 m to 10 km running, cross-country skiing, rowing, criterium racing, sprint and Olympic-distance triathlon. It is the classic answer for an athlete who has plateaued on a threshold-heavy diet — that is what Stöggl's and Muñoz's studies actually demonstrate. It also works as a late block rather than a whole season: the world-class triathlete Cejuela & Sellés-Pérez tracked across a 43-week macrocycle at 14.74±3.01 h/week was pyramidal in the early periodisation and only polarised at the end.

Not for you if

Do not prescribe polarised to an athlete below the floor (under ~6 h/week single-sport, under ~8 h/week triathlon) — the model degenerates into intervals with a bit of easy work attached. Do not make it the default for recreational athletes: the best current evidence points them toward pyramidal. Do not run it as the specific or race phase for a long-course triathlete — Muñoz et al. measured Ironman racing at 59±22% of race time between the thresholds, so a pure polarised build is specificity-poor at exactly the wrong moment; the distribution stays, the final block adds race-pace work. Same for marathon runners in the specific phase, since marathon pace sits in the middle zone for most athletes. Do not use it without a measured threshold anchor: if VT1/LT1 is guessed from an age-predicted HRmax formula the easy ceiling can be wrong by 10+ bpm and every number downstream is meaningless. Do not use it for athletes who cannot or will not execute genuinely easy sessions — ego pace on easy days, group rides, running with faster partners. Polarised executed badly is strictly worse than pyramidal executed well, because the easy volume loses its entire rationale. Avoid it in a fragile recovery state (heavy life stress, poor sleep, illness, a history of non-functional overreaching), in a first year of structured training, and for injury-prone runners if the whole 20% is placed in running — move that to the bike.

In a triathlon week

It transfers, but it has to be re-engineered rather than applied.

The swim will not polarise. Pool work is interval-structured by pool geometry and dominated by pace-clock and technique work; almost nothing in a normal set sits below VT1. Two consequences. Do not force the swim toward 80/20 — it destroys the session quality swimming actually needs. And exclude the swim from the whole-week distribution calculation, or report it in its own line: otherwise the overall number reads pyramidal and the athlete “fixes” it by deleting swim quality, which is the wrong fix. 80/20 Triathlon's workaround, if the swim must be counted, is to prescribe it at 75/25 by distance; Warden's stated reason is that most people swim in a fixed-distance pool, and that 75/25 by metres has empirically come out at 80/20 by time. It is a coaching approximation, not a validated equivalence.

Count hard sessions across the body, not per sport. Cap total genuinely hard sessions at two to three per week across all three disciplines, then decide which sport they land in. Percentages will never catch the seven-hard-session week; the session count will.

Put the intensity where it is cheapest. Cycling absorbs high intensity at the lowest mechanical cost and is also where easy volume is cheapest to accumulate. A defensible default for an injury-prone age-grouper: the bike carries the VO2max work, the run carries one quality session and a lot of easy volume, the swim carries its own naturally threshold-heavy structure. The run dose caps push the same way — at 36 km a week the run's above-threshold set is seven 400s, about nine minutes of work, so the bike has to supply the rest of the 30–40 min dose.

Per-discipline floors: roughly 4 h/week riding and 3 h/week running before calling a week polarised, which is what sets the ~8 h whole-week floor.

Race specificity breaks it for long course, and the evidence genuinely conflicts. Muñoz, Cejuela, Seiler, Larumbe & Esteve-Lanao (2014) followed nine Ironman athletes over 18 weeks at 68±14% Z1: more total time and more Z1 time correlated with faster racing (r=−0.69 and −0.92) while a higher percentage of Z2 time correlated with slower racing (r=0.94), and the authors specifically flag high-volume Z2 cycling as harmful. But Sellés-Pérez, Fernández-Sáez & Cejuela (2019), on recreational triathletes split into polarised (84.5/4.2/11.3) and pyramidal (77.9/18.8/3.3) groups before a half-Ironman, found middle-zone training time inversely correlated with swim and bike race time in the polarised group and with run and total race time in the pyramidal group — more middle, faster. The honest reading is that middle-zone volume should be periodised: small in base, substantial and race-pace-specific in the final 8–10 weeks. That is why Polarised pairs best with Blocks or Base → Build and hands off to race-specific work near the event rather than refusing it.

What the evidence says

Well supported. Elite endurance athletes across many sports accumulate the large majority of training below the first threshold — one of the most replicated observations in the field, from Seiler & Kjerland's original 11 skiers and 318 endurance sessions (75±3 / 8±3 / 17±4 by heart rate, 76±4 / 6±5 / 18±7 by session RPE, 92% agreement between the two methods, and 71% of all lactate measurements at or below 2.0 mmol/L) up to Muniz-Pumares et al. (2025) on 151,813 marathons from 119,452 runners. Some meaningful amount of work above easy is necessary: the low-intensity-only group in Seiler's 2013 interval study gained just 4.2%. Interval dose-response within the hard zone is unusually clean: 4 × 8 min at ~90% HRpeak beat both 4 × 4 at ~94% and 4 × 16 at ~88% in a direct randomised comparison. And threshold-dominant training in already-well-trained athletes produces little further gain — Stöggl's threshold group showed no significant improvement in VO2max, time to exhaustion or peak power.

Contested — do not assert polarised superiority as settled. Rosenblat et al. (2025), an individual-participant-data network meta-analysis of 13 studies and 348 athletes with Seiler as senior author, found no difference between polarised and pyramidal for VO2max (SMD −0.06, p=0.68) or time trial (SMD −0.05, p=0.34), and no significant difference between polarised and any other distribution; the one significant subgroup effect favoured pyramidal for recreational athletes (SMD −0.63, p<0.05). Silva Oliveira, Boppre & Fonseca (2024) found a small polarised advantage for VO2peak only (SMD 0.24, 95% CI 0.01–0.48, p=0.040), with time trial (p=0.92), time to exhaustion (p=0.24) and velocity/power at VT2/LT2 (p=0.75) all null, and the authors note that several included studies did not even report their distributions. Even in recreational runners, Muñoz, Seiler et al. (2014) found the 10-week difference between polarised (~77/3/20) and between-thresholds (~46/35/19) training was not statistically significant overall (5.0% vs 3.6% in 10 km, about 41 s apart); a polarised advantage appeared only in a per-protocol subset of the best-adhering runners.

Importantly, the two largest observational datasets point at pyramidal, not polarised. Muniz-Pumares et al. concluded that pyramidal was the most common approach, adopted by more than 80% of runners with the fastest marathon times, and that its prevalence increased in the fastest runners. Casado et al. (2022) concluded that highly trained and elite distance runners typically follow a pyramidal distribution, with polarised more characteristic of 1500 m runners than marathoners. Both support the narrow claim that the large majority of training sits below the first threshold; neither supports polarised over pyramidal.

Coaching lore, not evidence. The specific number 80/20 — Seiler calls it a population optimum, and observed elite values range from about 68% to 88% in the easy zone. “Zone 2 is junk miles” — elites spend 5–16% there, and Casado, Foster, Bakken & Tjelta (2023) describe world-class runners doing three to four threshold-region sessions per week at 2–4.5 mmol/L plus one VO2max session on 150–180 km/week, which sits directly against the 5%-of-sessions framing. “Polarised is scientifically proven best” — it is not. And polarised is a distribution, not a training system: it says nothing about progression, tapering, strength work or specificity, which is why it needs a periodisation overlay.

Provenance matters here. Seiler did not invent or prescribe 80/20; he measured it, and he is senior author on the 2025 meta-analysis that found no polarised advantage. The hard-edged community version is a coaching hardening of a descriptive observation.

How it goes wrong

Zone-number collision is the most common and most damaging error: an athlete reads “80% in zone 2” from a five-zone, Coggan or San Millán source, applies it to a three-zone chart, and spends 80% of their training in the exact zone the model says to minimise. Always name the model with the number, or name the physiological anchor instead.

Session-versus-time conflation. Prescribing 80/20 by session and then judging compliance from a device's time-in-zone chart. Correctly executed, this week reads about 86 easy / 8 in the middle band and at threshold / 6 above the second threshold in minutes, and Seiler's own athletes measured 91 / 6 / 2.6; an athlete “correcting” that chart adds a third hard session and overtrains. Seiler and Tønnessen also warn that heart-rate software overestimates time at low intensity and underestimates time at the highest workloads, so the chart is not even a clean record of the week. Related: applying Treff's Polarisation Index to session counts, which it is not defined on — it is defined on time or distance.

Warm-up and cool-down accounting. A 4 × 8 session is a hard session by intent but is 65–75% easy by time. Pick one convention and use it for prescription and reporting alike.

Treating 20% as a target rather than a ceiling. Two hard sessions is the prescription, three is a loading block, four is a mistake.

Lengthening the reps and keeping the watts. A four-minute band is not a six- or eight-minute band: 5 × 6 min at 115% FTP is half an hour above FTP and cannot be completed. Longer reps sit at the bottom of the range, or are ridden self-paced as the hardest effort sustainable across the set.

Making the hard zone maximal. Seiler's own data has 4 × 8 at 90% HRpeak beating 4 × 4 at 94–95%. Norwegian elite practice is explicitly controlled rather than exhaustive, with reps in reserve and a slight progressive build. The internet version is too hard on the hard days and therefore too hard on the easy days too.

Letting easy drift up. If the easy 80% creeps toward 84–86% HRmax the athlete is running a threshold-dominant programme while believing they are polarised.

Counting hard sessions per sport instead of per body — two hard bikes, two hard runs and three hard swim sets is seven hard sessions on one cardiovascular system.

Citing Stöggl & Sperlich (2014) as proof of 80/20. Its polarised group was 68/6/26 by time, and the groups were not volume-matched (POL 104±20 h, HVT 102±11 h, THR 84±7 h, HIIT 66±1 h over nine weeks), which weakens the causal claim badly.

Confusing Seiler's polarised training with Matt Fitzgerald and David Warden's 80/20 Endurance system. That is a different, time-based, seven-zone product whose plans deliberately include threshold and steady-state work, and whose result is pyramidal, not polarised. If an athlete wants that, they want Balanced.

Sources: sportsci.org europepmc.org europepmc.org europepmc.org pmc.ncbi.nlm.nih.gov pmc.ncbi.nlm.nih.gov pmc.ncbi.nlm.nih.gov europepmc.org pubmed.ncbi.nlm.nih.gov pubmed.ncbi.nlm.nih.gov pubmed.ncbi.nlm.nih.gov pmc.ncbi.nlm.nih.gov europepmc.org europepmc.org europepmc.org europepmc.org umh1617.umh.es scientifictriathlon.com 8020endurance.com