What the evidence actually shows
The question has been looked at properly. A Cochrane review pooled 34 randomised or quasi-randomised trials covering close to four thousand healthy infants under six months, and tested massage against a long list of outcomes.
On first pass the results look encouraging: weight, length, head circumference, hours slept over twenty-four, time spent crying, blood bilirubin and rates of diarrhoea all came out in favour of massage. Then the quality of the studies was taken into account. Twenty of the thirty-four were rated at high risk of bias, mostly for inadequate randomisation, unconcealed allocation or unblinded assessors. Once those were set aside, only length remained statistically significant. Every other significant finding disappeared either on that exclusion or at later follow-up.
Cortisol, night sleep, upper respiratory illness, anaemia, infant temperament, mental development and — notably — parent-infant interaction showed no effect at all. The reviewers' conclusion was that the evidence does not support the use of infant massage with low-risk parents and babies, and that future research would be better aimed at higher-risk groups, where a measurable effect is more plausible.
That is a thin result for a practice this widely recommended, and it is worth stating plainly rather than burying. It is not evidence that massage is harmful or pointless. It is evidence that the specific health benefits attached to it in most writing on the subject have not been demonstrated in healthy, full-term babies.
Where the evidence is stronger: babies born early
The picture is different on a neonatal unit, which is where most of the serious research has been done and where massage is used as a clinical intervention rather than a domestic one. NHS neonatal guidance cites weight gain, improved feeding and digestion, steadier heart rate, oxygen saturation and temperature, lower bilirubin, reduced stress and pain, and shorter length of stay.
It also gates the practice carefully. One protocol in use requires a weight above 1,200 g, at least 32 weeks corrected gestational age and demonstrated medical stability within the previous eight hours; the International Association of Infant Massage sets its minimum at term equivalent, above 37 weeks, and holds that massage should not be used below that age. Parents are the preferred people to deliver it, taught by a certified instructor on the unit.
The practical implication for a baby who was born early and is now at home is that the hospital is the place to ask. A unit that has taught a parent the sequence has already judged the baby ready for it; one that has not may have a reason.
Oil, and why the advice changed
This is the part of the subject where older writing is not merely dated but wrong, and it is worth being precise about.
Two oils were recommended for decades: olive oil and sunflower oil, both kitchen staples and both assumed to be harmless. A trial at Saint Mary's Hospital in Manchester put that assumption to the test, randomising 115 newborns to olive oil, sunflower oil or no oil at all, applied twice daily for twenty-eight days. Both oil groups ended the month with better-hydrated skin and with a measurably delayed development of the skin barrier compared with the babies who had nothing applied. The researchers judged that the hydration did not justify the possible cost to the skin's lipid layer, and declined to recommend either oil for healthy UK babies.
Olive oil carries a second mark against it. It is high in oleic acid, and a controlled study on adult forearms found that four weeks of it reduced the integrity of the outer skin layer and produced mild redness, while sunflower oil preserved the barrier. Current NHS neonatal guidance lists olive oil among the oils to avoid, alongside mineral or paraffin oils and anything scented or containing essential oils.
Putting that together with general NHS advice — plain water is best for a baby's skin in the first month, and lotions and oils are best avoided until at least a month old — produces a straightforward position:
- In the first month, no oil. Massage with bare warm hands, or over a vest, works perfectly well. The stroking is the point; the oil is only there to reduce friction.
- After a month, an unscented, plain, cold-pressed plant oil, used sparingly. Not olive oil. Not baby oil of the mineral or paraffin kind. Nothing perfumed, and nothing with essential oils added, whatever the label claims they do.
- A patch test first, on a small area, a day ahead — more so if there is eczema in the family.
- Nut-derived oils warrant extra thought where there is any family history of allergy, and are worth raising with a health visitor rather than deciding alone.
Washing and general skin care are a larger subject, set out separately under baby skin care.
When not to massage
Most of these come straight from clinical guidance and several are not obvious.
- Any fever or active infection. This is an absolute stop, not a judgement call.
- Within three days of an immunisation. Widely listed as a contraindication and almost never mentioned in popular accounts.
- A bleeding disorder, or any painful or contagious skin condition.
- Straight after a full feed. Waiting is the easier option for everyone.
- When the baby is asleep, hungry or crying. A massage is not something to administer to an unwilling baby; a crying baby wants feeding, holding or changing, and will not learn to associate the routine with calm if it arrives in the middle of distress.
- Below term-equivalent age, or in any baby who is medically unstable, except under the guidance of the unit caring for them.
During a session, the signals to stop are a baby who turns the head away, stiffens, splays the fingers, goes suddenly still and quiet, hiccups, or whose skin mottles or changes colour. Those are the recognised stress cues, and the correct response to them is to stop rather than to persevere gently.
How it is done
The room wants to be warm and free of draughts, because a baby with little clothing on loses heat quickly and has limited means of generating it. Hands should be warm before starting. Everything needed — towel, clean nappy, oil if the baby is over a month — is easier gathered beforehand than fetched halfway through.
The conventional sequence begins at the legs and feet, which are the least intrusive area and the one most babies accept first, and moves to the tummy, chest, arms and hands, and then the back with the baby turned over. Strokes are slow and light, running along a limb rather than pressing into it.
Two things that appear in older instructions should be dropped. Firm pressure on the head is one: the skull plates have not fused and the soft spots are exactly that, so the head is stroked only very lightly or left alone. The other is any instruction to squeeze or knead muscles to release knots. A baby does not have muscle knots. That sentence has been carried across from adult massage writing, where it belongs, and applied to an infant, where the pressure it implies is simply too much.
Ten to fifteen minutes is plenty, and finishing early because the baby has had enough is a success rather than a failure.
Colic, and what is actually advised for it
Relief of colic, wind and constipation is the most frequently repeated claim for baby massage, and it is not supported. The NHS guidance on colic lists comforting cuddles, holding the baby upright during feeds and winding afterwards, gentle rocking, a warm bath and soft background noise. Massage is not among them.
The same guidance is pointed about what to avoid: spinal manipulation and cranial osteopathy, on the grounds that there is little evidence they work and they may hurt the baby, and anti-colic drops, herbal remedies and probiotic supplements, on the grounds that there is no evidence they help. Colic resolves on its own by three or four months in almost every case.
What can honestly be said is that a warm room, an unhurried adult and fifteen minutes of calm handling may well help a fractious evening, and may help the adult as much as the baby. That is a modest claim and a defensible one. Presenting massage as a treatment for colic is neither.
Classes
Baby massage is taught through NCT courses, by instructors certified with the International Association of Infant Massage, and in many areas through health visiting teams and children's centres, where it is often free. A class teaches a sequence that is easy enough to learn from a demonstration and awkward to learn from a page, which is a fair reason to go.
The other reason is the one nobody puts on the poster. A weekly hour in a room of people at the same stage, with somewhere to ask the small questions that do not warrant an appointment, is worth a good deal to a new parent. The Cochrane reviewers found no measurable effect on parent-infant interaction, and still suggested that the groups most likely to benefit are those under the most strain. That is where to look for the value, rather than in the baby's weight chart.
The reasonable position
Baby massage is pleasant, safe when the few real cautions are observed, and free. It does not require a list of health benefits to justify it, and the list usually supplied does not stand up. A baby who enjoys being stroked will enjoy being stroked; one who does not, after a fair trial, is entitled to be left alone, and nothing is being lost by stopping.
Where it fits naturally is at the end of the day, as part of an unhurried wind-down, which is the context in which it does its most useful work. More on that under safer sleep for babies and babies' sleep in the first year.