REF-1314 · Rasmussen et al., 2013 · International Journal of Sports Physical Therapy · PubMed · Free full text

How much should I run each week before a marathon?

In 662 finishers, those averaging under 30 km a week had about twice the post-race injury risk of those on 30-60 km (RR 2.02). Retrospective, finishers only. B.

Main caveat: Retrospective self-report of weekly volume (an average for an undefined pre-race period) and injury, up to 37 days after the race; recall bias.
limited evidenceMeasured on recreational to trained athletes

Rasmussen and colleagues asked whether the risk of injury falls as weekly running volume before a marathon rises. Finishers of the H.C. Andersen Marathon in Odense (18 September 2011) were invited at the finish line and by e-mail to an online questionnaire open from 1 hour to 37 days after the race.

Of 2029 entrants, 680 answered (33.5%); 17 non-finishers and one runner whose name and time did not match the results were excluded, leaving 662.

The outcome was a new running-related injury after the marathon, severe enough to cause or be expected to cause a reduction in running distance, speed, duration or frequency for at least 14 days; the exposure was self-reported average weekly volume before the race in three bands. Sixty-eight runners (10.3%) reported an injury.

Crude, the risk was 2.34 times higher under 30 km a week than at 30-60 km; adjusted for previous injury and previous marathons it was 2.02 (95% CI 1.26-3.24), while over 60 km a week did not differ from 30-60 (1.13, 0.45-2.79).

Previous injury (adjusted RR 2.23) raised and a previous marathon (0.60) lowered the risk; younger age (under 35) went with more injury in the crude analysis; sex, BMI and the longest run in the last 6 weeks did not. The authors suggest at least 30 km a week before a marathon.

Our reading: a plausible volume floor for the marathon build, from a retrospective survey of finishers that could adjust for little and cannot show cause; the outcome is the race's toll, not injury during training.

Protocol

Setting and who: H.C. Andersen Marathon, Odense, Denmark, 18 September 2011; inclusion: completed the marathon and could read and write Danish; informed consent; n = 662 included of 680 questionnaires from 2029 entrants (return rate 33.5%).
Data collection: online questionnaire developed and pilot-tested by the investigators at three Danish marathons a year earlier; available 1 hour after the race; e-mail on day 9, reminder on day 19, closed on day 37; all relevant questions mandatory.
Outcome: a running-related injury (muscles, tendons, joints and/or bones, caused by running) after the marathon that caused or was expected to cause a reduction in distance, speed, duration or frequency of running for at least 14 days; muscle soreness, blisters and cramps excluded; location recorded.
Exposure: average weekly running volume before the marathon: under 30, 30-60 (reference) and over 60 km/week; also previous injury in the past 12 months, age, previous marathons, sex, BMI and the longest run in the 6 weeks before the race.
Analysis: univariate and multivariate binomial regression; a priori confounders (previous injuries, age, previous marathons, sex, BMI, longest run), but with one parameter per 15 injuries only previous injury and previous marathons could be adjusted for; STATA 11.2; no ethics approval required under Danish law for this observational design.
Funding and interests: no funding source and no conflict-of-interest statement appear in the article; the acknowledgements thank a medical student for the pilot work.

Key Results

ExposureCrude RR (95% CI)Adjusted RR (95% CI)
Under 30 vs 30-60 km/week2.34 (1.47-3.73), p < 0.012.02 (1.26-3.24), p < 0.01
Over 60 vs 30-60 km/week1.00 (0.40-2.50), p = 0.991.13 (0.45-2.79), p = 0.80
Previous injury in the past 12 months (yes vs no)2.30 (1.45-3.66)2.23 (1.41-3.52)
Previous marathon (yes vs no)0.46 (0.29-0.72)0.60 (0.37-0.95), p = 0.03
Age under 35 vs 35-50 years2.31 (1.44-3.73)not adjusted
Age over 50 vs 35-50 years0.82 (0.38-1.75)not adjusted
Longest run in the last 6 weeks: under 25 vs over 30 km1.27 (0.71-2.26), p = 0.42not adjusted
Women vs men0.77 (0.46-1.31)not adjusted
Who: 535 men and 127 women; age 41.4 ± 10.4 years; BMI 23.0 ± 2.3; previous injury in the past 12 months 273 of 662; running-club members 265.
Injuries: 68 of 662 (10.3%) — knee 22 (32.4%), foot and ankle 22 (32.4%), lower leg 12 (17.6%), thigh/hamstring 4, lower back 3, hip 2, Achilles tendon 1, missing 2.
Not reported: how many runners were in each weekly-volume band, so absolute risks per band cannot be reconstructed.
Numbers that do not add up:
The Results text gives the crude RR for under 30 km/week as 2.34 [1.47; 2.73]; Table 2 gives [1.47; 3.73]. Only the table's interval is symmetric on the log scale around 2.34 (2.34²/1.47 ≈ 3.72, our arithmetic).
The Discussion says the risk "increased to 134%"; an RR of 2.34 is an increase by 134%.
The Results describe runners over 60 km/week as also having "an increased risk" (RR = 1.13 [0.5; 2.8], p = 0.80), an interval that includes no effect.
Table 2's BMI row is labelled "(> 25 vs ≥ 25)", which cannot be the comparison made.

Mechanisms

Measured: an association between lower pre-race weekly volume and injury reported after the race, attenuated from 2.34 to 2.02 by adjustment for previous injury and previous marathons.
Authors' interpretation: a lower threshold for weekly volume, rather than only an upper one, should be considered; the relation between volume and injury is a balance between overuse and under-conditioning; experienced runners may have adapted and may read their bodies better.
Inference: for a runner averaging under 30 km a week, 42.2 km on race day is a single-session spike far above the recent longest run — the exposure the base links to overuse injury in REF-1102; the study did not measure that ratio, and its own longest-run variable (under 25 vs over 30 km in the last 6 weeks) was not associated with injury (crude RR 1.27, 0.71-2.26).
Inference: low volume may also mark less experience, less time or a disrupted build (an earlier injury); with two confounders adjusted for and age left out, part of the association may be who runs little, not running little.

Practical Application

How much should I run each week before a marathon? In 662 finishers, those averaging under 30 km a week had about twice the post-race injury risk of those on 30-60 km (RR 2.02). Retrospective, finishers only. B.
Use a weekly-volume floor before race-specific marathon work (B, observational; the floor as a gate is practice): the authors suggest at least 30 km a week; the base's own practice: "reach and hold about 30 km a week for several weeks before marathon-specific work", and grow the long run without single-run spikes (REF-1102, SYNTH-64).
More than 60 km a week showed no evidence of further protection here (adjusted RR 1.13, 0.45-2.79, wide; band sizes not reported): for a recreational marathoner the floor matters more than a high ceiling.
A short build changes the goal, not the race-day dose (inference): if the athlete cannot reach a repeatable base, consider a later or shorter race rather than cramming volume — now a line of SYNTH-64 ("if that is out of reach, a later or shorter race is the honest option", inference), not of SYNTH-90 or SYNTH-42; the nearest there is SYNTH-42's consistency gate (fewer than ~6 of the last 8 weeks with the usual session count: regularity first; practice).
Previous injury and a first marathon each carried risk (adjusted RR 2.23; a previous marathon 0.60): the first-timer with a recent injury and a thin build is the profile to slow down.

Limitations

Retrospective self-report of weekly volume (an average for an undefined pre-race period) and injury, up to 37 days after the race; recall bias.
Selection: finishers only (26.0% of entrants did not complete, some perhaps injured), a 33.5% response, possibly skewed either way (authors).
Outcome timing: an injury after the marathon, not during the build; recurrent injuries excluded.
Confounding: adjusted for two variables only; age, sex, BMI and experience beyond previous marathons were not in the adjusted model.
Reporting: group sizes per volume band absent; one confidence interval mistyped in the text; three wording errors (see Key Results).
Funding and interests: none stated.
How this entry was checked and changed
Why this level

an observational cohort over 200 sits in the rubric's B+ band; held at B because training volume and injury were both self-reported after the race, only finishers were included and only 33.5% of entrants answered, the outcome is an injury after the race rather than during the build, the model could adjust for only two confounders, and the number of runners in each volume band is not reported. Read from the full text (PubMed Central)

Full text (checked 2026-10-27)
From the 2029 persons signed up for the marathon, 680 questionnaires were gathered (return rate = 33.5%); 662 runners were included in the study (full text)
Sixty-eight runners (10.3%) reported an injury after the marathon; the injury had to cause or be expected to cause a reduction in running for at least 14 days (full text)
Adjusted for previous injury and previous marathons: under 30 km/week RR 2.02 [1.26; 3.24] vs 30-60 km/week; above 60 km/week RR 1.13 [0.45 ; 2.79], P=0.80 (full text)
Crude RR for under 30 km/week: 2.34 with [1.47; 3.73] in Table 2 and 2.34 [1.47; 2.73] in the Results text (full text)
Adjusted previous injury RR 2.23 [1.41; 3.52]; previous marathon 0.60 [0.37; 0.95] (full text)
Male/female 535/127; age 41.4 ±10.4 years; previous marathons yes/no 478 / 184 (full text)
Below 35 years RR 2.31 [1.44; 3.73]; longest training distance below 25 km RR 1.27 [0.71; 2.26] (full text)