How the best marathoners actually split easy and hard running
The argument over intensity distribution has two familiar camps. One says the best endurance athletes train "polarized": lots of easy running, a smaller dose of hard work, and very little in between. The other points out that marathoners seem to spend a fair amount of time at threshold, the uncomfortable middle that polarized models tell you to avoid. The research on highly trained and elite distance runners gives a reasonably clear answer for marathoners, and it is more nuanced than either camp's slogan.
First, some perspective. For the vast majority of runners, the big lever is running more, consistently, over months. How that running is split between easy, threshold and hard intensities is a second-order detail. It can matter at the margin for runners who already train steadily, and it needs a solid base of regular running behind it before it pays off.
Key takeaways
Most training models in this area use three zones: below the first lactate or ventilatory threshold (zone 1), between the two thresholds (zone 2), and above the second threshold (zone 3).
A polarized distribution puts most training in zone 1, a meaningful share in zone 3 and little in zone 2. A pyramidal distribution also puts most training in zone 1, with less in zone 2 and less again in zone 3.
A 2022 systematic review of highly trained and elite distance runners found they typically train pyramidal, and that marathoners leaned more pyramidal while 1500 m runners leaned more polarized.
The same review describes a shift from pyramidal training in the preparatory periods towards polarized training in the competitive period.
Both patterns share the same foundation: the large majority of training is easy. The debate is about how the remaining hard work is divided.
Intervention studies comparing polarized and threshold-heavy training are small and short, and they were run on already well-trained athletes.
The three-zone model
Most of the research uses a three-zone model anchored to two physiological markers. Zone 1 sits at or below the first lactate or ventilatory threshold, roughly where blood lactate starts to rise above resting levels (often around 2 mmol·L⁻¹). Zone 2 runs between the two thresholds, up to around 4 mmol·L⁻¹ or the second ventilatory threshold. Zone 3 is everything above that, the territory of VO2max intervals and shorter, faster reps.
The popular "Zone 2" of five-zone heart-rate models describes easy aerobic running, which falls in zone 1 here. Mixing the two up is a common source of confusion.
With three zones defined, the two distributions are easy to describe. Polarized training has zone 1 as the largest share, zone 3 as the second largest and zone 2 as the smallest. Pyramidal training has the shares falling in order: zone 1 largest, then zone 2, then zone 3.
Where the polarized idea came from
Stephen Seiler's work did much to popularise the polarized model. In a 2006 study, Seiler and Kjerland tracked 11 well-trained junior cross-country skiers over 32 days and 318 endurance sessions. By heart rate, about 75% of sessions fell in zone 1, 8% in zone 2 and 17% in zone 3, and session RPE and blood lactate gave a similar picture. They concluded that elite endurance athletes "train surprisingly little at the lactate threshold intensity".
Seiler's 2010 review in the International Journal of Sports Physiology and Performance widened the lens. Across descriptive studies of nationally and internationally competitive endurance athletes, about 80% of training sessions were performed at low intensity, with about 20% dominated by high-intensity work. That is the origin of the "80/20" rule of thumb. The review also found no convincing evidence that adding more high-intensity training to the programmes of already well-trained athletes gives long-term gains.
Two details are worth keeping in mind. The 2006 data come from skiers, and the 80/20 split in the 2010 review counts sessions. Counting sessions, minutes or kilometres can give different-looking splits from the same week, because a single interval session contains warm-up and recovery at low intensity.
A fixed 80/20 ratio is also a population average from highly trained athletes. kaizen works from the running you actually do: it accounts for the overall intensity of each run, from its pace and elevation, to set your training target for the week, and it recalculates as the week's running comes in.
What the research on distance runners shows
The most useful summary for runners is the 2022 systematic review by Casado, González-Mohíno, González-Ravé and Foster in the International Journal of Sports Physiology and Performance. It pooled ten studies of highly trained and elite middle- and long-distance runners and looked at intensity distribution, volume, periodization and training methods.
Its main findings:
Pyramidal is the typical pattern. Training volume fell from zone 1 to zone 2 to zone 3.
Both kinds of quality session appear every week. Elite runners used continuous tempo runs or longer intervals around the second threshold at least once a week, and shorter, faster intervals in zone 3 at least once a week as well. Adding more of the first type tilts a programme towards pyramidal, and adding more of the second tilts it towards polarized.
Event distance matters. Marathoners leaned more pyramidal. 1500 m runners leaned more polarized.
The distribution changes through the season. The authors recommend a traditional periodization pattern on a hard day–easy day basis, shifting from pyramidal training in the preparatory and precompetitive periods towards polarized training in the competitive period.
For a marathoner, the practical reading is that threshold work has a legitimate place. The event is run close to the threshold region, and the elite runners in these studies spent a real share of their hard training there. Polarized training remains a sensible pattern too, particularly closer to racing.
What the intervention studies add
Descriptive studies show what good athletes do. Testing which distribution causes better results takes trials that assign athletes to different programmes.
In a 2014 study, Stöggl and Sperlich randomised 48 well-trained runners, cyclists, triathletes and cross-country skiers to nine weeks of polarized, threshold, high-intensity or high-volume training. The polarized group showed the largest gains in VO2peak, time to exhaustion and peak speed or power. The threshold and high-volume groups showed little further change in most measures.
A 2019 meta-analysis by Rosenblat, Perrotta and Vicenzino pooled randomised trials comparing polarized and threshold models. Only four studies met the criteria and three made it into the meta-analysis. The pooled result showed a moderate effect favouring polarized training for time-trial performance, and the included studies scored 4 or 5 out of 10 on the PEDro quality scale.
These trials point in the same direction, and they are small, short and mixed across sports. None was a marathon build. They compare a polarized programme with one built mainly around threshold work, and the pyramidal pattern elite marathoners use sits between those two.
How much threshold work belongs in a marathon block
The research supports a few general statements for marathoners:
Most of the running should stay easy, in the region below the first threshold. Every distribution in this literature agrees on that.
Threshold-region work, such as tempo runs and longer intervals, is a normal part of an elite marathon build and featured in elite programmes at least weekly.
Faster zone 3 sessions also featured weekly, and the review suggests their share can grow as racing approaches.
The literature gives no precise number of weekly threshold sessions for a recreational marathoner, and published plans that set one are making a judgment for a generic runner. kaizen is hyper-personalised: it reads the intensity of the running you've actually completed and sets a weekly target from it, so a week with an extra tempo run and a week of mostly easy miles each count for what they were. You still choose the sessions, whether that's a club workout, a tempo run or an easy long run.
If you want to check your own pattern, the polarization index proposed by Treff and colleagues in 2019 is one tool. It combines the shares of zone 1, 2 and 3 into a single number, with values above 2.0 indicating a polarized distribution. Its value depends on how the zones were measured.
Sources and how sure we can be
The descriptions of the three-zone model and the two distributions, and the findings of the Casado review, come from the published abstract of that review. The Seiler and Kjerland figures, the Seiler 2010 summary, the Stöggl and Sperlich results and the Rosenblat meta-analysis come from those papers' abstracts. We have not read every full text.
There are real limits. The descriptive data on elite runners cover small numbers of athletes, and zone boundaries were set differently across studies, by lactate, ventilatory thresholds or heart rate. Counting by sessions, time or distance changes the percentages. The randomised trials are short and include several sports. Whether any of this carries over directly to recreational marathoners on five or six runs a week is an open question. Our reading that threshold work deserves a regular place in a marathon block is an interpretation of the elite data.
The split sits on top of the running
Polarized and pyramidal are two ways of dividing the hard part of a week. Both rest on a large volume of easy running, and that base is where nearly all runners find most of their improvement: more running, done week after week, at a level you can absorb. Once that is in place, deciding how much threshold work to include is a useful refinement.
See your training add up in kaizen
If you're months into a marathon build and care how your easy and hard running balance out, kaizen calculates your fitness from your recent running and past races, then sets the weekly training target your goal requires. Every run counts by its pace and elevation, and the target recalculates as the week goes. You decide which sessions fill it. Open kaizen on iOS or Android and see your fitness.
