Do you have to sleep train your baby?
No.
That is the complete answer. The longer version explains why the question even needs asking.
Throughout most of human history — and in most of the world today — infant sleep has been managed through proximity and responsive parenting, not formal behavioural training. Babies have been held to sleep, fed to sleep, and gradually supported toward independent sleep as their brains matured. The idea that there is a specific behavioural intervention that must be applied within a specific window, or sleep will go permanently wrong, is a culturally constructed, commercially reinforced, relatively recent invention.
It is not a biological requirement. It is not a clinical necessity. And the evidence does not show that babies who are not formally sleep trained have worse sleep, worse development, or worse wellbeing as a result.
For the research and the full picture, see our guide to cry-it-out. For the commercial context behind the pressure, see The Sleep Training Industry.
What happens if you don’t
The most important thing first: your baby will still learn to sleep.
Sleep consolidation — the gradual development of longer stretches, fewer wakes, and the capacity to settle without parental help — is a developmental process. It happens as the brain matures. The prefrontal cortex develops regulatory capacity. Circadian rhythms consolidate. Night waking reduces. This happens across a wide range of parenting approaches, and it is driven primarily by neurology, not by what the parent did at bedtime.
The timeline varies. Some babies sleep through reliably at six months; others have fragmented nights at 12, 18, or 24 months, and both are within the normal range. The most rigorous long-term follow-up available found no difference in sleep outcomes between sleep-trained and untrained babies by the time the children were five years old [1]. Both groups got there. They arrived via different routes and on different schedules.
You may be more tired in the short term. This is real and deserves acknowledgement. Responsive night parenting is not easy. Fragmented sleep sustained over months is a genuine challenge. The choice not to sleep train has a cost, and it is unfair to pretend otherwise. Some families will weigh that cost against the distress of extinction-based methods and conclude that sleep training is worth it for them. That calculation is valid.
Your baby’s attachment will not be harmed. Responsive night parenting — going to your baby when they call, feeding them if hungry, settling with contact — is consistently associated with secure attachment in the research. You are not building a problematic dependency. You are building a relationship. The evidence does not show that untrained babies are less securely attached than trained ones.
No longitudinal evidence shows worse outcomes for untrained children. Not in sleep. Not in behaviour. Not in emotional development. Not in cognitive outcomes. If failing to sleep train led to measurably worse results in any of these areas, that evidence would exist by now. It doesn’t [1].
Why you’re being told otherwise
If the evidence is this equivocal, why is sleep training presented as near-universal clinical advice?
Western cultural norms value early infant independence in ways that aren’t biologically universal. Cross-cultural and historical research shows that in most of human history and the majority of current non-Western cultures, co-sleeping, bed-sharing, and responsive night parenting are standard practice — not exceptions, not problems, not things that require apology [3]. The conviction that babies should sleep alone, through the night, from an early age is a specific cultural product of post-industrial Western societies. It is not biological wisdom.
Professional recommendations often outpace the evidence. When a paediatrician recommends sleep training, it can feel like a medical instruction backed by clinical consensus. But most paediatric guidance on sleep training is based on a limited evidence base interpreted through cultural assumption. The 2006 AASM review is frequently cited. The failure rates, objective measure gaps, and absence of long-term benefit are less frequently mentioned. A recommendation is not the same as evidence.
Survivorship bias shapes the social information you receive. You hear about the families for whom sleep training worked quickly and completely — the baby settled within three nights, the parents got four consecutive hours, everyone feels human again. You hear this story often, confidently, in playgroups and WhatsApp chats. You hear less often, and less confidently, about the families for whom it failed, the babies who escalated, the parents who tried and stopped and felt terrible about it. The anecdotal evidence available to you is skewed toward success [2].
Inadequate leave policies displace a systemic failure onto parenting choices. When a parent must return to work at six weeks, or twelve weeks, or even six months, the impossibility of functioning on fragmented sleep becomes acute. In that context, sleep training can feel not just desirable but necessary for survival. But the necessity is structural, not biological. The problem is inadequate parental leave, not an untrained baby. Sleep training becomes a coping mechanism for a system that hasn’t fixed the underlying problem.
Your other options
Wait and support. Continue responding to your baby at night in whatever way works — feeding, rocking, contact sleeping — and trust that development is doing the work. Most of what parents are trying to engineer in infancy arrives naturally through neurological maturation. The timeline varies. It is not predictable in advance. But it comes. For most of human history, this was the whole plan.
Optimise without training. Many sleep problems respond to non-behavioural changes: adjusting nap timing or duration, moving bedtime earlier or later to better align with the baby’s natural rhythm, improving the sleep environment (darker room, white noise, consistent temperature), and addressing any medical issues — reflux, ear infections, tongue tie — that might be disrupting sleep. It is worth ruling these out before reaching for a behavioural programme.
Gentle approaches. Bedtime fading — adjusting bedtime to match the baby’s natural sleep onset, then gradually shifting it earlier — can substantially reduce bedtime resistance without any distress component. Gradual retreat reduces parental presence progressively over time. Responsive settling maintains parental presence while gradually reducing the intensity of active help. The evidence base for these approaches is smaller than for extinction methods, but it exists and it is growing [4]. Blunden et al. (2022) found that responsive settling produced comparable sleep improvement with substantially less distress [4].
Share the load. If sleep deprivation is the acute crisis, the answer may not be training the baby — it may be restructuring how night care is shared. A partner taking alternate nights, a bottle of expressed milk enabling someone else to do one feed, a relative staying for a period, a postnatal doula. Reducing the load on one person can make an unsustainable situation sustainable without any change to the baby’s sleep.
None of these options guarantees a sleeping-through baby at a particular age. Neither does sleep training — its real-world failure rate is 40–50% [2]. The difference is that these options don’t require you to do something that feels wrong to you, and they don’t require your baby to reach escalating distress as part of the process.
The bottom line
You don’t have to sleep train. You can if you want to. Both choices are valid, and both paths lead to the same destination: a child who sleeps. The only genuinely wrong thing is being told you have no choice.
If you’re sitting with the question of whether to stop an attempt that isn’t working — or to not start one at all — read If Cry-It-Out Doesn’t Feel Right, Trust Your Instincts.
References below.