Question: which sports are truly risky? Are our intuitions — CrossFit is dangerous, running is safe — actually supported by evidence? Comparing sports requires a shared unit of measurement and a few methodological guardrails. Here is what the data say.
Opening: intuition vs data
Picture a novice runner preparing for their first half-marathon. Friends warn them about CrossFit: « it wrecks your joints. » They assume running, gentle and low-impact, is the sensible choice. But is that intuition actually correct?
Comparing injury risk across sports is harder than it looks. A rugby player and a trail runner don’t spend the same number of hours practicing, don’t compete under the same demands, and don’t report injuries through the same channels. Raw incident counts are meaningless without exposure data. Once you standardize by hours of practice, the risk hierarchy looks very different from what common sense suggests — and several results are genuinely counterintuitive.
This article synthesizes recent epidemiological data using the metric endorsed by the IOC, FIFA, and World Rugby: injuries per 1,000 hours of practice (IR/1000h). It covers methodology, a 12-sport comparison table, three counterintuitive lessons, and practical guidance for clinicians and coaches.
Why measure injuries per 1,000 hours of practice?
The injury incidence rate (IR) expressed per 1,000 hours of athletic exposure is the international standard recommended by the IOC 2020 consensus statement (Bahr et al., BJSM 2020), the FIFA consensus, and the World Rugby injury surveillance guidelines. The logic is straightforward: to compare two sports fairly, you must account for how much time athletes actually spend exposed to risk. A sport with 100 injuries across 10,000 hours carries exactly the same per-hour risk as one with 10 injuries across 1,000 hours — even though the raw counts differ tenfold.
Two injury definitions are in common use. « Time-loss » injuries — those causing at least one day of absence from training or competition — are the most widely adopted, since they represent a consistent and clinically meaningful threshold. The broader « medical attention » definition captures all injuries requiring any clinical contact, which inflates incidence figures and complicates cross-study comparisons.
One structural finding recurs across virtually every team sport with adequate surveillance: match (competition) risk is typically 5 to 10 times higher than training risk. In elite men’s football, the match IR is 16.3/1000h versus 1.8/1000h in training. In professional rugby, the gap is even more extreme — over 80/1000h in matches versus roughly 2.6/1000h in training. This asymmetry has direct implications for prevention strategy.
Summary table — injury risk by sport
| Sport | Incidence (/1000h) | Context |
|---|---|---|
| Professional rugby (match) | 83–115 | Super Rugby, 5 seasons — Schwellnus et al., BJSM 2018 |
| Walking football 40+ (match) | 37.6 | Match IR — Price et al., Sports 2025 |
| Senior women’s handball (match) | 27.7 | Slovenian senior league — Barić et al., 2021 |
| Elite women’s football (match) | 18.4 | UEFA Women’s ECIS, 4 seasons — Hallén et al., BJSM 2024 |
| Novice runners | 17.8 | Meta-analysis — Videbæk et al., Sports Med 2015 |
| Elite men’s football (match) | 16.3 | EURO 2024 / Copa América 2024 — Verhagen et al., BMJ Open SEM 2025 |
| Recreational runners | 7.7 | Meta-analysis — Videbæk et al., Sports Med 2015 |
| CrossFit (median) | ~3 | Systematic review, 25 studies, n=12,079 — Rodríguez et al., 2020 |
| Recreational functional fitness | 1.22 | Brazilian cohort — Utida et al., Human Movement 2024 |
| Beach tennis (recreational/pro) | 1.11 | Brazilian cohort, n=235 — Laras et al., 2025 |
| Recreational alpine skiing | <0.5 | University courses — Kalkusová, Studia Sportiva 2018 |
| Climbing / bouldering (courses) | <0.5 | University courses — Kalkusová, Studia Sportiva 2018 |
Quick read — three counterintuitive findings from this table. First, walking football — widely perceived as a low-risk adaptation of soccer for older adults — produces a match IR of 37.6/1000h, higher than elite women’s football in competition. Second, CrossFit’s median incidence (~3/1000h) is less than half the recreational running figure (7.7/1000h) and nearly six times lower than novice running. Third, recreational alpine skiing sits below 0.5/1000h — lower than any team sport training context in this table.
Three counterintuitive lessons
The CrossFit-is-dangerous myth
CrossFit has a reputation as a high-injury discipline, driven partly by online anecdotes and partly by early studies with methodological limitations. The systematic review by Rodríguez et al. (2020), covering 25 studies and 12,079 participants, found an IR range of 0.2–18.9/1000h with a median of approximately 3/1000h. Gardiner et al. (2020) reported a narrower range of 0.27–3.3/1000h in their brief review, and Feito et al. (2020) found that non-competitive CrossFit participants had IRs between 0.39 and 1.30/1000h. The most affected body regions are the shoulder (26%), spine (24%), and knee (18%).
Placed in context, CrossFit’s median risk is lower than recreational running and far lower than any team sport practiced in competition. The wide range in the literature reflects heterogeneity in study populations (competitive vs recreational), injury definitions, and surveillance quality — not an inherently catastrophic activity. For a clinician or coach, the message is clear: CrossFit warrants attention to technique and progressive loading, not categorical avoidance.
Novice running: an underestimated risk
Running is generally regarded as a safe, accessible activity. The landmark meta-analysis by Videbæk et al. (Sports Med 2015) complicates that image significantly. Novice runners — those in their first months of regular practice — sustain injuries at a rate of 17.8/1000h (95% CI: 16.7–19.1). That figure is higher than elite men’s football in competition. Recreational runners with established training habits fare considerably better at 7.7/1000h, and competitive track athletes (middle and long distance) drop to around 2.5/1000h.
The mechanism is well understood: rapid increases in training load without adequate musculoskeletal adaptation drive the majority of running injuries, most of which are overuse in nature. The clinical and coaching implication is straightforward — progressive overload protocols, adequate recovery, and structured onboarding for novice runners are not optional extras; they are the primary prevention lever. Framing running as inherently « safe » delays the intervention that beginning runners most need.
The competition vs. training gap
In most team sports, the absolute majority of practice hours are spent in training, yet the injury risk per hour concentrates overwhelmingly in competition. In professional rugby, Schwellnus et al. (BJSM 2018) recorded match IRs of 83–115/1000h against a training IR of approximately 2.6/1000h (West et al., 2019) — a ratio exceeding 30:1. In elite men’s football, the match IR (16.3/1000h) is roughly nine times the training IR (1.8/1000h).
This asymmetry matters for prevention resource allocation. If a team plays 30 matches per season at 90 minutes each, that represents 45 hours of high-risk exposure — yet prevention programs often focus on training-period interventions (warm-up protocols, strength programs) rather than match-day preparation and return-to-play protocols. The data suggest that targeting the peri-competition window — including return-to-play decisions, travel fatigue, match congestion, and pre-match warm-up — offers the highest leverage for reducing actual injury burden.
For clinicians and coaches: methodological cautions
Cross-sport comparisons using IR/1000h are valuable but carry inherent limitations that responsible interpretation requires acknowledging.
Injury definition heterogeneity. Studies using the « medical attention » definition (any clinical contact) report higher IRs than those using the « time-loss » threshold (≥1 day absence). The IOC 2020 consensus (Bahr et al.) and the FIFA/Fuller 2006 football consensus both recommend time-loss as the primary metric — cross-study comparisons that mix definitions are not directly valid.
Sex, age, and competitive level. IRs differ meaningfully between elite and recreational athletes, between sexes (women’s handball match 27.7 vs men’s 10.6; women’s football 18.4 vs men’s 16.3), and across age groups. Figures from professional cohorts should not be applied uncritically to recreational populations.
Surveillance quality. Embedded medical staff capture more injuries than self-report questionnaires — counterintuitively, higher-quality surveillance yields higher apparent incidence. When comparing studies, consider the surveillance method as carefully as the injury definition.
For practical prevention: think in exposure hours, not raw counts; target resources toward high-risk periods (the novice phase, the peri-competition window, return-to-play); and use the IR/1000h framework when communicating risk to athletes or patients — it replaces intuition with numbers.
Key takeaways
- IR/1000h is the international standard for injury incidence comparison across sports (IOC 2020 consensus, Bahr et al., BJSM).
- Professional rugby in competition (83–115/1000h) and elite football in competition (16–18/1000h) dominate the top of the risk hierarchy among the sports reviewed here.
- CrossFit, recreational alpine skiing, and climbing carry low overall risk per hour of practice (<3/1000h) — despite their reputations.
- Novice running is an underestimated risk at 17.8/1000h — higher than many team sport competition IRs and far above recreational running (7.7/1000h).
- Inter-sport comparisons require caution: injury definition, competitive level, sex, and surveillance quality all affect the numbers. Always read the fine print.
Full bibliography (download)
The detailed bibliography, including complete references by discipline (football, rugby, handball, running, CrossFit, winter sports, paddle sports, tennis, adolescent comparisons) and methodological consensus statements, is available as a PDF (French content with bilingual references):
Key references
Bahr R, Clarsen B, Derman W, et al., 2020. International Olympic Committee consensus statement: methods for recording and reporting of epidemiological data on injury and illness in sport 2020. British Journal of Sports Medicine. Link
Videbæk S, Bueno AM, Nielsen RØ, Rasmussen S., 2015. Incidence of running-related injuries per 1000 h of running in different types of runners: a systematic review and meta-analysis. Sports Medicine. Link
Schwellnus M, Derman W, Jordaan E, et al., 2018. Elite athletes travelling to international destinations >5 time zone differences from their home country have a 2-3 fold increased risk of illness. British Journal of Sports Medicine — Super Rugby injury surveillance. Link
Rodríguez MA, García-Calleja P, Terrados N, et al., 2020. Injury in CrossFit®: a systematic review of epidemiology and risk factors. The Physician and Sportsmedicine. Link
Hallén A, Tomás R, Ekstrand J, et al., 2024. UEFA Women’s Elite Club Injury Study: 1527 injuries over four seasons. British Journal of Sports Medicine. Link
Verhagen E, Okholm Kryger K, Bahtijaragević Z, et al., 2025. Injury incidence at UEFA EURO 2024 and CONMEBOL Copa America 2024. BMJ Open Sport & Exercise Medicine. Link
Cet article est aussi disponible en français — read this article in French: Risque de blessure selon le sport.
NutriCellScience, Mark DOWN — EN edition
Laisser un commentaire