REF-1196 · Mujika & Padilla, 2003 · Medicine & Science in Sports & Exercise · Paper · PubMed

How much should I cut in a taper?

Keep the intensity, cut the hours a lot (this review allows up to 60-90%), drop sessions by no more than 20%, and expect about 3%. An expert review, not a pooled analysis.

B+moderate evidenceMeasured on a mixed group

Mujika and Padilla define the taper as a progressive nonlinear reduction of the training load over a variable period, meant to remove the fatigue of daily training without losing its adaptations. Their abstract sets out how: keep training intensity, reduce training volume by up to 60-90%, and reduce training frequency only slightly, by no more than 20%.

The optimal duration "ranges between 4 and more than 28 d"; progressive nonlinear tapers are stated to be more beneficial than step tapers; and performance usually improves by about 3% (usual range 0.5-6.0%), which the authors attribute to cardiorespiratory, metabolic, hematological, hormonal, neuromuscular and psychological changes.

Our reading: this is a widely used statement of the taper recipe from the leading taper group, and its ranges are wider than the later pooled optimum (REF-107: 2 weeks, 41-60% volume, frequency unchanged).

A 4-to-28-plus-day "optimum" is a span, not a recommendation; the volume ceiling of 90% and the 20% frequency allowance come from a narrative reading of mostly small studies, not from a test of where benefit stops.

Protocol

Design: narrative review. Search method, number of studies, inclusion criteria and study populations: not given in the abstract.
n = not given in the abstract (no participant or study count is stated).
Full text: not read (Lippincott, not open access; no repository copy found in OpenAlex).

Key Results

What the abstract states (no effect sizes or confidence intervals are given):

Taper variableStatement in the abstract
Intensitymaintained
Volumereduced, "up to 60-90%"
Frequencyreduced slightly, "no more than 20%"
Durationoptimal duration "between 4 and more than 28 d"
Shapeprogressive nonlinear tapers more beneficial than step tapers
Expected gainabout 3% (usual range 0.5-6.0%)
Attributed causescardiorespiratory, metabolic, hematological, hormonal, neuromuscular and psychological changes
Aim, in the authors' words of the abstract: minimise accumulated fatigue without compromising adaptations.
Not in the abstract: how each range was derived, which sports and levels it rests on, whether the shape comparison is from controlled studies, and any statement about detraining risk or a particular week-by-week schedule.

Mechanisms

Authors' interpretation: the performance gain comes from positive changes across cardiorespiratory, metabolic, hematological, hormonal, neuromuscular and psychological status (listed in the abstract, no data given there).
Inference: the recipe treats intensity as the stimulus that preserves adaptation and volume as the main source of fatigue; that is the logic the direct trials in this batch test — high intensity kept versus removed (REF-1199) and frequency kept versus cut (REF-1200).

Practical Application

How much should I cut in a taper? Keep the intensity, cut the hours a lot (this review allows up to 60-90%), drop sessions by no more than 20%, and expect about 3%. An expert review, not a pooled analysis.
The volume cut may go past the pooled optimum, but the pooled data do not back it. The later meta-analysis put the best results at 41-60% (REF-107, A), and in REF-3492 the ≥60% band was not significant (SMD −0.21 [−0.58, 0.16]); this review allows up to 60-90%, and in short direct trials runners improved with interval training cut 80% (REF-1200, daily-training group) or continuous running cut 60% (REF-1201).Neither source shows where the benefit stops (inference from the two).
Frequency: trim, do not empty the week. A cut of up to 20% is within this review's recipe; the one trial that removed every third day of a 6-day taper saw no significant gain in 4 runners, while daily training gave a significant within-group gain in 5; the groups were not compared with each other (REF-1200, B). The base keeps frequency, or trims it by no more than about a fifth (SYNTH-34).
Duration is a range, not a number. 4 to more than 28 days in this review; the pooled tapers ran 6 to 28 days, and only the ≥22-day band (N = 10) was not significant (REF-3492). The base's three-week ceiling is narrower than this review allows (see Cross-References).
What to expect: about 3% in this review's summary (0.5-6.0%); the later meta-analysis's mean gain was smaller, 1.96% as Spilsbury et al. cite it (REF-1201, full text; the base's REF-107 entry gives effect sizes, not this figure). Promise a better race, not a large one.

Limitations

From the abstract only. The ranges cannot be traced to the studies behind them; the full text was not read.
Narrative, not systematic. No search strategy, no pooling and no risk-of-bias appraisal is described; the ranges are the authors' synthesis.
A range is not an optimum. "Between 4 and more than 28 d" covers almost every taper ever tested; "up to 60-90%" gives an upper bound with no lower bound or dose-response.
Dated. Written before the two taper meta-analyses (REF-107, 2007; REF-3492, 2023), which narrow the volume band and do not support a frequency cut by default; where they disagree, the pooled analyses carry more weight for the direction, and this review for the width of what has been seen to work.
Conflicts and funding: not stated in the abstract.
How this entry was checked and changed
Why this level

authoritative narrative review from the leading taper group; the abstract describes no systematic search and no pooled analysis, and the full text was not available (not open access), so which studies stand behind each range is not checked