Double Threshold Training: Norwegian Method and the Amateur Version

Double threshold training means running two lactate-controlled threshold interval sessions on the same day, usually six to eight hours apart (Bakken, 2026). The format became known through the Ingebrigtsen brothers, who averaged 140 to 160 km (87 to 99 miles) a week with 23 to 25% of that volume at or above anaerobic threshold pace (Tjelta, 2019). The part most explainers leave out: no controlled study has tested whether the model works (Casado and colleagues, 2023).

The lactate, mileage and session numbers published by the method's creators show what it is and where the evidence stops. What transfers to runners on 50 to 90 km (31 to 56 miles) a week is the single-session version, which is how the two Norwegian plans in the catalog for the half marathon and the 10K are built.

Key takeaways

• Double threshold: two threshold sessions on the same day, 6 to 8 hours apart, at 2 to 3 mmol/L of lactate, inside 150 to 180 km (93 to 112 miles) a week (Bakken, 2026; Casado and colleagues, 2023).
• No controlled trial of the method exists. The most direct experiment compared one long session with two short ones in 14 athletes and measured acute responses only; its authors note that the literature on the format is limited (Talsnes and colleagues, 2024).
• The model has been described only in 1500 m and 5000 m runners; for the marathon its applicability "remains uncertain" (Casado and colleagues, 2023).
• On 50 to 90 km (31 to 56 miles) a week, what transfers is the single session: three sub-threshold workouts a week with 1-minute rests, truly easy runs and an easy long run.
• Without a lactate meter, sub-threshold sits near half marathon pace, at 83 to 87% of maximum heart rate, with enough breath for short sentences (Haugen and colleagues, 2022; Bakken, 2026).

What double threshold training is

Double threshold started as one runner's answer to his own plateau. Marius Bakken, a Norwegian with a 13:06.39 best over 5000 m who ran the 2000 and 2004 Olympics, writes that his results stagnated between 1996 and 1998 (Bakken, "About Marius"; Bakken, "The Norwegian Model", 2022). From 1998, Norwegian runners began using lactate meters to control running intensity. From 1999 onward, the threshold sessions were doubled.

The central change was a lower lactate ceiling. Instead of the standard 4.0 mmol/L, Bakken ran his intervals below 3.0 mmol/L, "usually staying from 2.3 up to 3.0" (Bakken, 2022). On his 2026 page he calls that 2.3 to 3.0 mmol/L range the "Golden Zone" and says he ran more than 5,500 lactate tests to calibrate the method (Bakken, 2026). On the 2022 page he also warns that the text is "by no means a scientific article".

The current definition is plain: "two controlled threshold workouts are done on the same day, usually separated by about six to eight hours". Beyond eight to ten hours, Bakken says, you are approaching two separate workouts rather than one structured block (Bakken, 2026).

The Ingebrigtsen numbers

The reference case is the three Ingebrigtsen brothers, all European 1500 m champions. According to the abstract of Tjelta (2019), they averaged 140 to 160 km (87 to 99 miles) a week, with 23 to 25% at and above anaerobic threshold pace, and lactate was measured during all interval sessions. The study is observational: it describes the training, it does not test the method.

Casado and colleagues (2023), citing Tjelta, give the session detail: up to four threshold sessions a week, with reps of 2000 to 3000 m or 400 to 1000 m. Recoveries run from 20 seconds to 1.5 minutes, and each session totals 8 to 12 km (5 to 7.5 miles). A fifth weekly session goes up to zone 4 or 5 (the example given is 20×200 m uphill), and the longest run does not exceed 21 km (13 miles).

Who deserves the credit is contested. Bakken writes that between 2009 and 2013 Henrik Ingebrigtsen's father asked him for session-by-session advice (Bakken, 2022). Gjert Ingebrigtsen, according to journalist Alex Hutchinson, "has minimized the importance of Bakken's influence" (Hutchinson, Outside, 2023). For a runner, who invented it matters less than what has been published with numbers.

What a double threshold day (and week) looks like

A double threshold day pairs long intervals in the morning with short ones in the afternoon, both in the same lactate range. In Bakken's description, the morning holds reps of 6 to 10 minutes, such as 2000 m or 3000 m; the afternoon holds 400 m to 1000 m reps, about 15 to 20 seconds per kilometer faster (about 24 to 32 seconds per mile) (Bakken, 2026). For reps of 1 minute or less, he considers very short recoveries of 15 to 30 seconds ideal (Bakken, 2022).

Bakken's own week, February 2004

The week Bakken published from his own training shows how the doubles sit among everything else (Bakken, 2025):

• Monday, Wednesday, Friday: two easy runs a day, zone 1.
• Tuesday: AM, 6×2000 m (lactate 2 to 3 mmol/L); PM, 20 to 25×400 m (lactate close to 3.0).
• Thursday: AM, 4×10 min (lactate 2 to 3 mmol/L); PM, 10 to 12×1000 m (lactate close to 3.0).
• Saturday: AM, 10×200 m in 26 to 28 seconds.
• Sunday: long run, "never longer than 16 to 18 km" (10 to 11 miles).
• Total: 180 km (112 miles) for the week.

Two things stand out for anyone coming from marathon plans. The long run is short for the total mileage, and the only fast session of the week is ten 200 m strides. Almost everything else is easy running. The sample week in Casado and colleagues (2023) follows the same design, with doubles on Tuesday and Thursday. The long intervals sit at 2.5 mmol/L (5×6 min with 1 min rest); the short ones at 3.5 mmol/L (10×1000 m with 1 min and 25×400 m with 30 s). Saturday brings 20×200 m hill sprints, and the total comes to about 110 miles (177 km), mostly easy running, as Hutchinson (2023) describes it.

Why split threshold work into two sessions

The reasoning, in the words of the model's own authors, is to accumulate more threshold volume at a lower cost. Casado and colleagues (2023) argue that the interval format allows faster speeds than continuous runs in the same lactate range. The conditions are short reps, so lactate does not rise progressively, and recoveries long enough for it to return to levels near the first lactate threshold. Tight control of lactate, in their view, "might be advantageous".

Staying below threshold rather than above it is about recovery. The paper cites a lab study by Burnley and colleagues, using isometric quadriceps contractions, in which exercise above critical torque generated a rate of global and peripheral fatigue four to five times greater than exercise below it. It is not running data, but it illustrates the cost of crossing the threshold. Bakken, according to the same paper, "could perform a much higher training volume" at threshold than when he ran zone 4 intervals, without the elevated catecholamines of that zone (Casado and colleagues, 2023).

The paper offers no specific physiological mechanism for placing both sessions on the same day. The justification is organizational: cluster the load on hard days, keep easy days truly easy, and fit more threshold into the week. Bakken adds hypotheses of his own, such as the "muscle tone" from the first session priming the second, and presents them as experience rather than evidence (Bakken, 2026).

The experiment: one long session or two short ones

Talsnes and colleagues (2024) compared the two formats head to head. The authors note that there is limited scientific literature to support double threshold training. Fourteen male endurance athletes (11 cross-country skiers and 3 runners, VO2max 69.2 ± 4.2 mL/kg/min) completed, in counterbalanced order, one 6×10 min session (SINGLE) and two 3×10 min sessions separated by 6.5 hours (DOUBLE). External intensity was identical: 90% of each athlete's speed at 4 mmol/L of lactate.

With the same external load, the split format ended the day with a lower resting heart rate, lower perceived exertion and less fatigue and soreness the next morning. Bakken's summary: "Same volume. Less cost." (Bakken, 2026). The authors, however, measured acute responses only: no performance outcome, no training week. They write that "future training intervention studies are required" and note that the participants were not familiar with double threshold training before the study (Talsnes and colleagues, 2024).

What the science still does not know

No controlled trial has tested whether double threshold works better than other ways of organizing the same volume. The line comes from the model's own authors: "no controlled studies have tested the efficacy of this training model", which rests "on the basis of only observational studies and reports" (Casado and colleagues, 2023). The paper is a position piece signed by Casado, Foster, Bakken and Tjelta, that is, by the method's creator and by the researcher who described the Ingebrigtsens.

The same paper acknowledges four limits:

• The training characteristics and their effects "have been described only in 1500 m and 5000 m runners".
• Applicability "in other endurance events, such as the marathon, remains uncertain".
• The model "also may represent an increased risk of injury/overtraining syndrome".
• The values that mark the zones "should be detected for each athlete through physiological tests".

The evidence base is also small. The systematic review by Kelemen and colleagues (2023) found only seven articles with systematically recorded training, covering 13 elite runners in total. Hutchinson sums up the state of play: the studies needed "are really hard studies to do", and he does not "expect that we'll get definitive answers anytime soon" (Hutchinson, 2023).

Tjelta himself urges caution before transferring the Ingebrigtsen case. "They've been lucky with their genes, they must have been," he told ScienceNorway; "you can try to follow their exercise programme, but there's no hope of getting just as good" (Schjølberg, ScienceNorway, 2019). Haugen and colleagues (2022) make the general version of the same point: "Many unsuccessful athletes have applied the same 'recipe' as successful runners."

Who can actually run doubles

Double threshold days make sense in weeks of 150 to 180 km (93 to 112 miles). Bakken writes that 120 km (75 miles) is "a minimum" for middle and long distance for many, and that 160 to 180 km (99 to 112 miles) "for most will work the best". For 5000 m runners he gives a range of 150 to 220 km (93 to 137 miles) (Bakken, 2022). Casado and colleagues describe the model as low-intensity running filling a total of 150 to 180 km a week (Casado and colleagues, 2023).

Two double days, with a full easy day between them

Inside that volume, the dose is two double days a week. The original model "involved twice-a-week two threshold sessions", plus one weekly session with a different, slightly higher-intensity stimulus (Bakken, 2022). After testing three hard days, Bakken concluded that "on a general basis, two threshold sessions are sufficient". In 2026 he adds that, for most trained runners doing two double days a week, at least one full easy day between them "is not optional" (Bakken, 2026).

He tested the excess himself. "I tried numerous variants with threshold work every day, mostly sub-threshold," he writes, "leading fairly quickly to overtraining"; the daily back-to-back stimulus never allowed full recovery (Bakken, 2026). The 2022 page reaches the same verdict: the effect of doing threshold work every day "is little" and "the chance of overtraining is far greater". In summer, his doubles were "reduced to single threshold work, with only occasional doubles" (Bakken, 2025).

For a runner on 50 to 90 km (31 to 56 miles) a week, the math does not work. Two days with two threshold sessions each would eat most of the weekly mileage and leave little room for the easy running that makes up most of the original model. Haugen and colleagues (2022) note that most injuries are attributed to rapid and excessive increases in training load, and that under-performance from overtraining tends to go with high volumes. The realistic alternative is the single session.

The amateur version: single sub-threshold sessions

The part of the method that transfers to 50 to 90 km (31 to 56 miles) a week is the controlled threshold session, not the double day. Instead of two workouts on the same day, you run three sub-threshold sessions a week on separate days, with short rests, surrounded by truly easy runs and an easy long run. Online, the format picked up the nickname "Norwegian singles", which Outside Run describes as a "largely crowdsourced training system" that grew out of a forum (Outside Run, 2025). Treat the name as a label, not as evidence.

The structure itself, though, has support in elite practice and in correlational data. Haugen and colleagues (2022) refer to "threshold" as an intensity close to half marathon pace. The typical sessions they list are 10 to 12×1000 m with 1 min recovery and 6 to 8×1500 to 2000 m with 1 to 2 min, with a recommended total of 30 to 75 min for elite runners.

The correlational data point the same way. Among 85 male runners, Casado and colleagues (2021) found that easy runs (r ≥ 0.68) and tempo runs (r ≥ 0.50) were the activities that accounted for significant variance in performance. The correlation with long-interval training was weak (r ≥ 0.22). It is a correlational study, not a trial.

How the catalog's Norwegian plans apply this

The two Norwegian plans in the Magic Training catalog use this design. The half marathon plan runs 14 weeks at 60 to 90 km (37 to 56 miles) a week over six running days, for runners already at 55 km (34 miles) or more. In the loading weeks (1 to 9) there are three sub-threshold sessions, all with 1 min rest: 5 to 8×6 min, 9 to 14×3 min and 5 to 6×1 km. The third session alternates 1 km, 2 km and 3 km reps. The biggest session, 8×6 min, adds up to 48 min at controlled threshold, inside the 30 to 75 min range Haugen and colleagues describe for elites.

Week 1 shows how it fits together:

• Monday: rest.
• Tuesday: 3 km easy + 5×6 min sub-threshold (1 min rest) + 2 km.
• Wednesday: 9 km (5.6 miles) easy.
• Thursday: 3 km + 10×3 min (1 min) + 2 km.
• Friday: 8 km (5 miles) easy.
• Saturday: 3 km + 5×1 km (1 min) + 2 km.
• Sunday: 16 km (10 miles) long run, easy.

Weeks 4 and 8 are lighter. From week 10 to 13, Saturday turns into half marathon pace (3×3 km with 2 min; 2×5 km with 3 min; 6 km continuous; 4 km), and the long run peaks at 24 km (15 miles) in week 10. Race week: 4×3 min sub-threshold on Tuesday, 3×2 min at race pace on Thursday, race on Sunday.

The 10K plan follows the same logic over 12 weeks, at 55 to 80 km (34 to 50 miles) a week on six running days, for runners already at 50 km (31 miles) or more. Week 1 has 4×6 min, 8×3 min and 4×1 km, all with 1 min rest, and a 14 km (8.7 miles) long run. VO2max work (5×1 km) only appears in week 9, and race-pace work in weeks 10 to 12.

Double threshold and single sessions side by side

How to hit the intensity without a lactate meter

Without a meter, you control sub-threshold intensity with pace, heart rate and the talk test, used together. Bakken himself acknowledges that "few runners have access to lactate testing in training" and calls heart rate "the most practical substitute, but it should not be used alone" (Bakken, 2026). Perceived effort, in his view, "is the least reliable of the three on its own".

In practice, start with pace. Take your paces from a recent race, never from a goal time, in the Magic pace calculator: sub-threshold sits close to half marathon pace. Cross-check with heart rate, using your maximum heart rate and your training zones in the 83 to 87% band. Remember that heart rate climbs through a session even at a constant pace: in Talsnes and colleagues (2024), it went from 168 to 173 bpm within the single session. Finally, the talk test: you should manage short sentences, not a chat.

Why chasing one lactate number is false precision

Even with a meter, the reading carries an error the size of the target zone. In Tanner and colleagues (2010), the Lactate Pro had a typical error of 0.5 mmol/L within the same device and the Lactate Scout 1.0 mmol/L. In direct comparison, the Scout read up to 0.9 mmol/L higher than the Pro.

Bonaventura and colleagues (2015) tested six hand-held analysers and found reliability generally below 0.5 mmol/L between 1 and 10 mmol/L. Even so, two units of the same brand (Lactate Pro) differed significantly in every band, and the authors conclude that "a prudent scientist would use the exact same unit". The sampling site also shifts the reading: in Zhong and colleagues (2025), during upper-body exercise on a SkiErg, fingertip lactate averaged 0.66 mmol/L higher than earlobe lactate.

Bakken's "Golden Zone" of 2.3 to 3.0 mmol/L is 0.7 mmol/L wide. Device and sampling errors of 0.5 to 1.0 mmol/L cover the whole zone. That is why both Bakken and Casado and colleagues insist that the values must be individualized. For most amateurs, pace, heart rate and the talk test used together are the realistic control.

Common mistakes in sub-threshold training

Running sub-threshold too fast

This is the mistake the method was built to prevent. Hutchinson reports that runners "used to hammering interval workouts really struggle to go easy enough" in these sessions (Hutchinson, 2023). Bakken's rule: "If lactate is drifting above 3 mmol/l by the final repeat, the intensity is too high" (Bakken, 2026). Without lactate, use half marathon pace as the ceiling and the talk test as the alarm.

Doing threshold every day

Bakken tried it and reports overtraining "fairly quickly" (Bakken, 2026). His conclusion for elites, inside 180 km (112 miles) a week, was two threshold days a week with at least one full easy day between them. In the catalog plans, that becomes three single sessions on alternating days; do not add a fourth.

Letting easy runs drift into "moderate"

In Bakken's model, easy running stays below 1.0 mmol/L of lactate and below 70% of maximum heart rate (Bakken, 2022). Elite runners perform 80% or more of their running volume at low intensity throughout the year (Haugen and colleagues, 2022). If your easy runs creep into Haugen's zone 2 (73 to 82% of maximum heart rate), they stop being recovery. The logic of the 80/20 split is a useful check on your week.

Copying elite mileage

The Ingebrigtsens' 140 to 160 km (87 to 99 miles) came after years of progression: Jakob "didn't do that until he was 17-18 years old," according to Tjelta (Schjølberg, ScienceNorway, 2019). Haugen and colleagues note that most injuries are attributed to rapid and excessive increases in training load, and that the younger the training age, the more careful the progression should be (Haugen and colleagues, 2022). A plan for 60 to 90 km (37 to 56 miles) is not a watered-down version of the method; it is the dose your mileage can carry.

Trusting a single lactate number

One isolated lactate reading does not tell you whether you are in the right zone: the device and the sampling site shift it by 0.5 to 1.0 mmol/L (Tanner and colleagues, 2010; Zhong and colleagues, 2025). Casado and colleagues recommend detecting each athlete's values through physiological tests, and Bakken writes that his values "also need to be individualized" (Casado and colleagues, 2023; Bakken, 2022). If you do measure, always use the same device at the same site, and compare trends rather than absolute numbers.

FAQ

Do I need a lactate meter to train with the Norwegian method?

Not for the single-session version. Bakken states that few runners have access to lactate testing and points to heart rate, combined with the talk test, as the practical substitute (Bakken, 2026). For elites, the meter is there to hold 2 to 3 mmol/L across 8 to 12 km (5 to 7.5 miles) of threshold running per session (Casado and colleagues, 2023). For an amateur, half marathon pace from the calculator and 83 to 87% of maximum heart rate do the job, and the meter itself carries an error of 0.5 to 1.0 mmol/L (Tanner and colleagues, 2010).

What pace and heart rate is sub-threshold?

Close to half marathon pace, at 83 to 87% of maximum heart rate and a perceived exertion of 14 to 16 on the Borg scale, which corresponds to 2.0 to 3.5 mmol/L of lactate (Haugen and colleagues, 2022). Bakken gives 80 to 87% for well-trained runners (Bakken, 2026). To turn that into a pace in min/mile or min/km, start from a recent race, never from a goal time.

Are two threshold sessions on the same day better than one long session?

In acute cost, yes; in performance, nobody knows. In the Talsnes experiment, two 3×10 min sessions 6.5 hours apart left 14 athletes with a lower recovery heart rate, lower perceived exertion and less fatigue the next morning than one 6×10 min session (Talsnes and colleagues, 2024). The study did not measure performance, and no controlled trial has tested the model (Casado and colleagues, 2023).

Does the Norwegian method work for the marathon?

Unknown. Casado and colleagues write that the model has been described only in 1500 m and 5000 m runners and that its applicability to the marathon "remains uncertain" (Casado and colleagues, 2023). In a systematic review of ten studies, elite marathoners generally followed a pyramidal distribution: most running easy, a slice at threshold, little above it (Casado and colleagues, 2022). That does not contradict the sub-threshold logic, but it does not test it either. For marathon training, methods with a longer track record remain the reference.

How many threshold sessions a week should an amateur run?

Three single sessions on alternating days, if you already run 50 to 55 km (31 to 34 miles) or more a week, as in the catalog's Norwegian plans. Below 50 km (31 miles) a week, run fewer quality sessions: three of them on low mileage leave little room for easy running, which makes up most of the kilometers in the original model (Bakken, 2025). Doubles are for runners at 120 km (75 miles) or more (Bakken, 2022).

Double threshold is a tool for 180 km (112 miles) a week, built by and for track runners, with no controlled trial behind it. What it teaches a runner on 50 to 90 km (31 to 56 miles) is simpler: three sub-threshold sessions a week, at the right pace, with short rests, and everything else truly easy. If you are not sure which plan fits your mileage and your race, the plan finder quiz sorts it out.

References

• Bakken M. The Norwegian Model. mariusbakken.com, 2022 (original description of the model). mariusbakken.com
• Bakken M. The Norwegian Model Revisited. mariusbakken.com, Kristiansand, 21 April 2025. mariusbakken.com
• Bakken M. Double Threshold Training. mariusbakken.com, Kristiansand, 22 March 2026. mariusbakken.com
• Bakken M. About Marius. mariusbakken.com. mariusbakken.com
• Tjelta LI. Three Norwegian brothers all European 1500 m champions: What is the secret? International Journal of Sports Science & Coaching, 2019;14(5):694-700. DOI 10.1177/1747954119872321
• Casado A, Foster C, Bakken M, Tjelta LI. Does Lactate-Guided Threshold Interval Training within a High-Volume Low-Intensity Approach Represent the "Next Step" in the Evolution of Distance Running Training? International Journal of Environmental Research and Public Health, 2023;20(5):3782. PMC10000870
• Talsnes RK, Torvik PØ, Skovereng K, Sandbakk Ø. Comparison of acute physiological responses between one long and two short sessions of moderate-intensity training in endurance athletes. Frontiers in Physiology, 2024;15:1428536. PMC11319183
• Kelemen B, Benczenleitner O, Tóth L. The Norwegian double-threshold method in distance running: Systematic literature review. Scientific Journal of Sport and Performance, 2023;3(1):38-46. DOI 10.55860/NBXV4075
• Haugen T, Sandbakk Ø, Seiler S, Tønnessen E. The Training Characteristics of World-Class Distance Runners: An Integration of Scientific Literature and Results-Proven Practice. Sports Medicine - Open, 2022;8:46. PMC8975965
• Casado A, González-Mohíno F, González-Ravé JM, Foster C. Training Periodization, Methods, Intensity Distribution, and Volume in Highly Trained and Elite Distance Runners: A Systematic Review. International Journal of Sports Physiology and Performance, 2022;17(6):820-833. PubMed 35418513
• Casado A, Hanley B, Santos-Concejero J, Ruiz-Pérez LM. World-Class Long-Distance Running Performances Are Best Predicted by Volume of Easy Runs and Deliberate Practice of Short-Interval and Tempo Runs. Journal of Strength and Conditioning Research, 2021;35(9):2525-2531. PubMed 31045681
• Tanner RK, Fuller KL, Ross ML. Evaluation of three portable blood lactate analysers: Lactate Pro, Lactate Scout and Lactate Plus. European Journal of Applied Physiology, 2010;109(3):551-559. PubMed 20145946
• Bonaventura JM, Sharpe K, Knight E, Fuller KL, Tanner RK, Gore CJ. Reliability and Accuracy of Six Hand-Held Blood Lactate Analysers. Journal of Sports Science and Medicine, 2015;14(1):203-214. PMC4306774
• Zhong F, Chen Z, Gu Z, Wang X, Holmberg HC, Li Y. Comparison of lactate measurements from earlobe and fingertip capillary blood using Biosen S-Line and lactate scout analyzers. European Journal of Applied Physiology, 2025;125(1):145-156. PubMed 39164441
• Hutchinson A. How the "Norwegian Method" Is Changing Endurance Training. Outside, 23 February 2023 (journalism). outsideonline.com
• Schjølberg UG. What is the secret to the Ingebrigtsen brothers' success? ScienceNorway, 1 October 2019 (journalism; English version by Nancy Bazilchuk). sciencenorway.no
• Outside Run. Finally, a Norwegian Method We Can All Try. Outside Run, updated 11 June 2025 (journalism; describes a forum-sourced system, cited here for the name only). run.outsideonline.com
• Data for the catalog's Norwegian plans extracted on 20 October 2026 from each plan's workout table, counting as "threshold" the blocks tagged in the plan's threshold zone.