REF-1293 · Tønnessen et al., 2020 · International Journal of Environmental Research and Public Health · Paper · PubMed · Free full text

Are more, shorter threshold sessions better than fewer long ones?

In 15 elite skiers, 12 weeks of 136 min/week as 2 long or 4 short sessions: only 2 long improved the TT within-group; between-group ES trivial (0.17). B-.

Main caveat: No between-group test. The paper compares within-group p values (p = 0.04 vs p = 0.16) and reports between-group effect sizes only.
limited, mixed evidenceMeasured on highly trained athletes

Tønnessen et al. tested whether two long threshold sessions a week beat four shorter ones at the same intensity and weekly volume in elite endurance athletes. Twenty national-level cross-country skiers and biathletes were randomised for 12 weeks of summer dry-land training.

The low-frequency group (LF) did 8 × 8 min with 2-min recoveries and 6 × 12 min with 3-min recoveries each week; the high-frequency group (HF) did 4 × 8 min twice and 3 × 12 min twice — 136 min a week in intensity zone 3 (82-87% HRmax, 2.5-4.0 mmol/L lactate) in both, all as roller-ski skating. Other training was easy, with at most one extra high-intensity session a week.

One LF and four HF athletes dropped out (illness 4, accident 1), leaving 9 and 6. In LF the 8-km uphill roller-ski time trial improved from 31.8 to 30.6 min (median, p = 0.04; seven of nine faster); in HF it did not change significantly (p = 0.16). The between-group effect size was trivial (0.17).

LF also improved %VO2max at 4 mmol/L lactate (80.7 → 85.4%), oxygen cost (41.0 → 39.6 mL/kg/min) and submaximal heart rate (144 → 134 beats/min), with between-group effect sizes of 0.52, 1.14 and 0.95; HF changed none of these significantly. VO2max did not change in either group.

The authors conclude that two longer sessions are superior and fit "best practice" in world-class athletes.

Our reading: the data show that splitting the same threshold work into twice as many sessions did not help, and hint that the longer sessions helped economy; with 9 vs 6 athletes, no between-group test and a trivial between-group effect on performance, they do not show that two beat four.

Protocol

Design: randomised parallel trial, 12 weeks, May to August, dry-land; randomisation method not described; groups matched at inclusion (n = 10 each); no blinding reported. Ethics: institutional committee; Norwegian centre for research data (26732).
Participants: 20 cross-country skiers and biathletes competing at national level for at least 3 years; n = 15 analysed — HF 5 men + 1 woman, LF 8 men + 1 woman. Dropouts: LF 1 (illness), HF 4 (3 illness, 1 accident); athletes not completing the whole period were excluded.Medians (IQR): age 22.0 in both; VO2max 67.8 (65.5-70.2) HF and 70.7 (67.0-73.9) LF mL/kg/min; training last 12 months 565 (431-638) and 650 (520-755) h; last 2 months 65 and 75 h (no significant differences).
Intervention (all roller-ski skating, intensity zone 3 = 82-87% HRmax, 2.5-4.0 mmol/L): LF — one session of 8 × 8 min (2-min recoveries, 64 min work) and one of 6 × 12 min (3-min recoveries, 72 min) a week, with at least one and typically two days between them. HF — two sessions of 4 × 8 min (32 min) and two of 3 × 12 min (36 min) a week; two of the four on consecutive days, one day off before the third. Weekly zone-3 work 136 min in both.
Other training: low intensity (< 82% HRmax) as usual, plus at most one high-intensity session (> 87% HRmax) a week; logged in an electronic diary; weekly volume, intensity distribution and strength training did not differ between groups. Intensity checked with lactate and HR in 1-2 group sessions a week for the first 4 weeks, then selected sessions.
Tests (before and after, 2-day taper, 48 h to 6 days apart): submaximal incremental roller-ski treadmill test (5-min stages, +2% incline, stopped above 4 mmol/L) for economy, HR and %VO2max at 4 mmol/L; VO2max test; 8-km uphill roller-ski time trial outdoors (30-35 min), same skis and poles, same start order.
Analysis: medians (IQR); Wilcoxon matched-pairs test within groups; Cohen's d for the difference between groups (0.2-0.6 small, 0.6-1.2 moderate); no between-group significance test.
Funding and interests: "This research was funded by the Norwegian Olympic Federation"; article processing charge paid by Kristiania University College; no conflicts of interest declared.

Key Results

Outcome (median, IQR)Two long sessions (LF, n = 9): pre → postFour short sessions (HF, n = 6)Between groups
8-km roller-ski time trial, min31.8 (30.3-35.0) → 30.6 (30.1-34.1), p = 0.04no significant change, p = 0.16ES 0.17 (trivial); no test
%VO2max at 4 mmol/L lactate80.7 (79.7-84.9) → 85.4 (79.9-88.2)%, p = 0.04p = 1.00ES 0.52 (small)
Submaximal VO2 (economy), mL/kg/min41.0 (40.3-42.8) → 39.6 (39.3-41.2), p = 0.01no significant changeES 1.14 (moderate)
Submaximal HR, beats/min144 (137-152) → 134 (124-144), p = 0.01no significant changeES 0.95 (moderate)
VO2maxno significant changeno significant change—
Completed the 12 weeks9 of 106 of 10—
Time trial: seven of nine LF athletes were faster after the 12 weeks; HF individual data are in a figure.
Compliance: all included athletes completed more than 80% of the interval sessions, with no difference between groups; skipped sessions were mainly at the athletes' request because they felt tired.
Training load: no difference in weekly volume, intensity distribution or strength training between groups during the intervention.
HF post-test values were not given in the text, only within-group p values; the between-group comparison rests on effect sizes.

Mechanisms

Measured: the long-session group became more economical (lower oxygen cost and heart rate at the same submaximal speed) and used a larger fraction of VO2max at 4 mmol/L; VO2max did not move in either group.
Authors' interpretation: a larger, more concentrated stimulus per session (64-72 vs 32-36 min in zone 3) may better drive peripheral adaptations — capillarisation through shear stress and muscle strain, mitochondria and aerobic enzymes — and greater glycogen use per session may add a signal for mitochondrial biogenesis; more days between threshold sessions may let athletes "absorb" the training.
Inference: the four-session group did two threshold sessions on consecutive days and lost more athletes to illness (3 vs 1); fatigue or illness may have diluted its response, and the remaining 6 are a selected subset.

Practical Application

Are more, shorter threshold sessions better than fewer long ones? In 15 elite skiers, 12 weeks of 136 min/week as 2 long or 4 short sessions: only 2 long improved the TT within-group; between-group ES trivial (0.17). B-.
Do not split a fixed dose of threshold work into more, shorter sessions to "spread the load" (B-): the four-session version showed no extra gain and lost more athletes.With REF-005 (three vVO2max sessions a week added nothing over one in highly trained runners) and REF-1295 (speed-endurance sessions every second day matched every fourth day for 10 km), it is the third small study in trained athletes in which more frequent quality sessions showed no extra gain; none is large or shows equivalence.
Long threshold sessions are a reasonable way to carry the week's zone-3 work (B-, economy findings, within-group): 64-72 min of work at 82-87% HRmax per session, as 8 × 8 or 6 × 12 min, improved economy in national-level skiers. Not from this paper; the base's own practice: for amateurs the length of threshold work follows SYNTH-26's quality-session framework and the interval shapes it points to.
Not a test of how many quality sessions to do: both groups did the same weekly dose; the count in SYNTH-26's decision table (usually two quality sessions for amateurs) stays practice.

Limitations

No between-group test. The paper compares within-group p values (p = 0.04 vs p = 0.16) and reports between-group effect sizes only; the time-trial effect was trivial (0.17). A significant change in one group and not the other is not evidence of a difference between them.
Small and unbalanced. 9 vs 6 analysed after 1 vs 4 dropouts; completers only; the authors acknowledge reduced power and did not assess adherence differences.
Baseline differences in training history. LF had higher median training volume in the past 12 months (650 vs 565 h) and 2 months (75 vs 65 h), and higher VO2max (70.7 vs 67.8), not significant in groups this small.
Spacing confound. HF did two threshold sessions on consecutive days; frequency and recovery between sessions changed together.
No session RPE recorded (the authors' limitation), so the perceived cost of the two formats is unknown.
Reporting. The economy speed is 10.8 km/h in the Methods and 10.9 km/h in the Results; HF post-test values are not given in the text.
Funding and interests: funded by the Norwegian Olympic Federation (a national sports body); no conflicts declared.
How this entry was checked and changed
Why this level

n = 15 analysed of 20 randomised, read from the full text (open access, CC BY). A parallel RCT under 20 is B; held at B- because no between-group significance test was run — the superiority conclusion rests on within-group p values while the between-group effect on the time trial was trivial (ES 0.17) — and dropout was unbalanced (4 of 10 vs 1 of 10), with completers only analysed

Full text (checked 2026-10-27)
Twenty elite cross-country skiers and biathletes; HF group n = 10 four interval sessions per week, LF group n = 10 two longer interval sessions; one drop-out in the LF group, four in the HF group, three due to illness (full text)
LF: 8 × 8 min intervals in I-zone 3 with 2 min recoveries and 6 × 12 min with 3 min recoveries; HF: twice-weekly 4 × 8 min and two sessions of 3 × 12 min; total prescribed weekly training time in I-zone 3 was 136 min (full text)
LF time-trial 30.6 (30.1–34.1) min vs. 31.8 (30.3–35.0) min, p = 0.04; HF p = 0.16; ES of LF training vs. HF training ES = 0.17 (full text)
% VO2max at anaerobic threshold LF 85.4 (79.9–88.2)% vs. 80.7 (79.7–84.9)%, p = 0.04; HF p = 1.00; ES = 0.52 (full text)
Submaximal oxygen consumption LF 39.6 (39.3–41.2) vs. 41.0 (40.3–42.8) mL/kg/min, p = 0.01, ES = 1.14; submaximal HR 134 (124–144) vs. 144 (137–152) bpm, p = 0.01 (full text)
A Wilcoxon matched-pairs signed rank test was used to test for differences between each parameter before and after the 12-week intervention period (full text)
This research was funded by the Norwegian Olympic Federation (full text)