welfare evidence library • question
Racing at two - what does the evidence actually show?
A six minute read - Asked after a yearling sale • Published August 2026
where this page starts
Racing two-year-olds is defended with a real argument, and it deserves stating properly.
The case goes like this. The horse is a precocial animal — born ready to run, at the mare's side and covering ground within days. Bone is not a fixed structure; it responds to load, and it responds best while it is still adapting. Introduce a young horse to work at the right time and you build a stronger skeleton than you would by waiting. On this reading, holding a horse back is not kindness — it is a missed window.
And there is research behind it. Large studies have found that horses who race at two go on to have longer careers and more starts than horses who begin at three or later (Velie et al., 2013; Tanner et al., 2013). Controlled work has found that structured exercise during growth produces measurable musculoskeletal benefits (Rogers et al., 2008). A review of the growth literature concluded that by two years old the horse has reached most of the standard markers of skeletal maturity (Rogers, Gee & Dittmer, 2021).
People who make this case are not inventing it. It is published, it is peer-reviewed, and it is repeated in good faith by vets and trainers who care about horses.
So this page doesn't start by asking whether two-year-old racing should stop.
It starts by taking the argument seriously and asking two questions of it:
Does the research show what it is said to show? And what does racing at two do to the horse?
first, how to read the research
This page spends a moment on the evidence itself before getting to it. That is not a dodge. On this question the research is the argument, so it is worth knowing three things about how it was built.
What was measured
Look closely at the headline findings and one thing stands out: the outcomes are career length, number of starts, and earnings.
Those are real measurements. They are also commercial measurements. A horse can have a long career, many starts, and good earnings while being in pain for much of it. Career length tells you the animal kept racing; it does not tell you how he felt while he did. Almost none of the large studies measured what the horse experienced — no pain scoring, no behaviour, no stress physiology, no follow-up on the horse's body after retirement.
So the finding is narrower than the way it gets used. "Racing at two does not shorten a racing career" is not the same claim as "racing at two does not harm a horse." The second is what the first is usually taken to mean.
So the finding is narrower than the way it gets used.
"Racing at two does not shorten a racing career" is not the same claim as "racing at two does not harm a horse."
The second is what the first is usually taken to mean.
Who was counted
This is the part that matters most, and it is not a criticism anyone in the field would find surprising.
To appear in a study of horses who "raced at two," a horse has to have made it to a race at two. Every horse who broke down in training, went lame, was found unsound, was sold on, or died before their first start is invisible in the data. They were never counted.
That is survivor bias, and here it is not a small effect. Follow a whole foal crop instead of the horses at the racetrack and the picture changes sharply. In one Australian cohort, 73% of registered foals entered training and 64% ever started a race — and only 14% had their first start as a two-year-old (Crawford et al., 2021). In a British and Irish cohort, of the two-year-olds who did make it into training, 61% raced at least once and just 5% earned enough to cover their training fees (Wilsher, Allen & Wood, 2006).
So the horses in the "raced at two" group are, by definition, the ones who stood up to it. Comparing them with horses who started later is not comparing two treatments. It is comparing survivors of one with a mixed group that includes the horses who couldn't. The researchers behind the Victorian cohort study said as much: working from subsets of the population, rather than the whole crop, introduces selection bias (Crawford et al., 2021).
There is a further twist. Age at first race start turns out to be strongly heritable — one New Zealand analysis put the heritability at around 0.5, meaning much of the variation between horses is genetic rather than managed (Chin et al., 2025). If precocity is substantially inherited, then "raced at two" is partly a marker of the horse a horse already was. Robust horses race early and last longer. The early racing needn't be doing the work.
Who paid, and who framed the question
Most of what we know about racehorse careers exists because racing paid to find out. The funders across this literature include racing boards, breeders' bodies and industry research corporations — in Australia, much of the foundational wastage work was funded through the Rural Industries Research and Development Corporation (now AgriFutures) (Bailey, 1998).
This is not a scandal. Industry funding is how most applied animal science gets done, and without it we would know far less than we do. But it is worth holding in mind, because the effect of funding on findings is itself well studied. Across nearly 4,600 studies in human medicine, industry-sponsored research was more likely to report results and conclusions favourable to the sponsor than independently funded research on the same questions (Lundh et al., 2017). Nobody claims fraud. The mechanism is subtler: which questions get asked, which outcomes get chosen, which findings get promoted.
You can see the framing effect without accusing anyone of anything. The influential 2021 review concluding that horses are skeletally mature by two declared no external funding — and its lead author was, at the time, editor-in-chief of the journal section in which it appeared, a conflict the paper itself declares alongside a note that he was blinded from review (Rogers, Gee & Dittmer, 2021). That is disclosed, above board, and common in a small field. It is also a picture of how small the field is.
The simplest observation is the one about the questions themselves. The literature asks: does racing at two shorten a career? It rarely asks: what is it like to be a two-year-old horse in race training? The first is a question about an asset. The second is a question about a horse. Only one of them has been well funded.
None of that makes the research wrong. It means the research answers a narrower question than the debate assumes, and that the gaps in it are not randomly placed. Read the rest of this page with that in mind.
PHYSICAL EVIDENCE
Does the whip work?
Does the whip make horses run faster?
When researchers timed races section by section, horses reached their highest speeds where the whip was not used. The whip was used most in the closing sections, when horses were already fatigued — that is, when they had least left to give (Evans & McGreevy, 2011).
Does it improve steering, reduce interference, or make racing safer?
Whip-free races (where jockeys hold the whip but don't use it) were compared with ordinary whip-permitted races across 126 races and 1,178 starters. There was no difference in steering, interference, incidents, or finishing times (Thompson et al., 2020).
On safety, the evidence goes further than "no benefit."
Whip use has been linked to horse falls in jumps racing (Pinchbeck et al., 2004) and to fatal leg fracture (Parkin et al., 2006).
If the whip were truly essential, the rules would agree with each other — and they don't.
Norway allows the whip only for genuine safety, not for encouragement. Japan and most US states set no numerical limit and leave it to the rider's discretion. Britain and Ireland sit in between, capping the number of strikes. A tool that were genuinely critical to a fair or safe race wouldn't be regulated in half a dozen contradictory ways across the racing world (see Table 1 and Table 2).
The distinction that matters
There is a sound reason for a rider to be able to correct a horse who is ducking or veering into danger — and the evidence doesn't argue with that. It's exactly why the countries that have moved away from the whip — Norway, Sweden, Denmark — still let riders carry one for genuine safety.
The line the evidence draws is between two different uses:
- the whip for safety correction — a narrow, defensible use, and
- the whip for encouragement — striking a horse to make them run faster or try harder.
Every finding on this page is about encouragement.
Encouragement is the practice with no measured benefit — and, as the rest of the page shows, a measurable cost to the horse...and the industry.
What the whip does to the horse's body
The padded whip is often assumed to protect the horse. Frame-by-frame analysis of race footage found this isn't reliable: a visible indentation appeared on the horse in 83% of strikes, and the unpadded section of the whip made contact 64% of the time (McGreevy et al., 2012). Padding doesn't dependably prevent contact or impact.
Table 1
Table 2
MENTAL-STATE EVIDENCE - what the horse experiences
What does the horse feel?
A common and reasonable-sounding argument is that a horse's skin is thick, so a padded whip can't really hurt them. It's worth testing directly.
Horse skin and human skin have the same density of nerve endings in the layer that detects pain, and the outer skin is no thicker where horses are struck (Tong et al., 2020). The physical basis for "they don't feel it the way we would" doesn't hold up.
Around that finding sits a settled scientific picture:
Horses are sentient and experience pain — they are aware of positive and negative feelings, pain among them (Cambridge Declaration 2012 ; NY Declaration on Animal Consciousness 2024).
Assessing pain in animals has been part of science for over 80 years (Broom, 2001).
A horse showing no obvious reaction is not evidence of no pain (Shutt 1987; Broom, 2001).
The equine pain face is a validated tool for recognising pain in horses (Gleerup 2015).
It's also worth noticing the language. Regulators describe striking a horse as "encouragement" — a word that carefully steps around what the horse is experiencing (Uldahl & Mellor, 2025)
Image provided by http://horseracingphoto.co.uk and is licensed under a Creative Commons Attribution 3.0 Unported License.
Image used with permission © Emma Preikschat
SOCIAL EVIDENCE - what a practice means
What does the public see?
Whipping has a long history as a tool of control and coercion — over enslaved people, and over animals as a means of domination (Wadiwel, 2013). That history is part of what the public reads into the image of a horse being struck.
A Tasmanian court recently ruled that padded whips,
When a horse is hit to win a race, it signals that the animal is treated as equipment rather than a partner (Luke et al., 2026). And representative surveys find that many — by some measures, most — people do not support whipping horses (World Horse Welfare / Voconiq, 2024; Wilson et al., 2021).
The fear that ending the whip would drive punters away isn't borne out. Few racing followers say they'd stop betting if the whip were banned (Wilson et al., 2021) — which matches the lived experience of Norway, Sweden and Denmark, where whip-free racing has carried on without losing its betting or its competitiveness.
"Does it (the whip) look good? Definitely not."
Where that leaves us
None of this tells racing what it should value — that isn't what evidence does.
What it can do is test the reasons the whip is usually defended on. On speed, steering, interference and safety, the studies here found no benefit from striking a horse for encouragement. On the horse's side of it, the research found contact and indentation on the horse's body, and an animal well equipped to feel pain from both.
Which leaves the question the sport keeps arriving at from every direction. Not whether racing could continue without the whip — Norway settled that decades ago — but why, when the whip doesn't do what we say it does, we still reach for it?