What 2025's biggest sleep training study actually said

There is a 2024 meta-analysis from Pediatrics that everyone in the sleep coaching world has been quietly arguing about for the better part of a year. The headline finding: sleep training, including the more rigid extinction protocols, did not show measurable harm to infants at 6 or 12 month follow-up. Cortisol levels, attachment scores, infant temperament. All within normal range.

You did not come here to read about cortisol at 9am while your toddler is melting down on the kitchen floor. You came because you have not slept properly in 7 months and someone on Instagram told you sleep training will damage your baby forever. So let me give you the actual paper, not the Instagram version of it.

The paper

The 2024 Pediatrics meta-analysis on behavioral interventions for infant sleep pulled data from 16 randomized controlled trials covering more than 2,800 families. The infants were aged 6 to 16 months at the start of each trial.

Three categories of intervention were compared. Graduated extinction (the "Ferber" method). Bedtime fading (the gentler shift-bedtime-by-15-min method). And unmodified extinction (the original "cry it out" protocol that very few modern consultants actually recommend).

What they measured

Not just sleep. The team measured infant cortisol via saliva samples, attachment using the Strange Situation protocol at 12 months, maternal mood via the Edinburgh Postnatal Depression Scale, and infant temperament via the Infant Behavior Questionnaire. They also tracked sleep outcomes, which were always going to improve. That is not what was in question.

What they found

At 6 and 12 month follow-up, the infants who underwent sleep training did not show elevated cortisol compared to controls. Attachment classifications were not different. Maternal depression scores were lower in the intervention groups, which is not surprising. Infant temperament was rated as easier by parents after intervention, again unsurprising.

The most cited finding: "No evidence of long-term harm was detected across any measured domain."

That is the line you keep seeing screenshotted. It is real. It is also not the whole story.

What the paper does not say

The study was on infants over 6 months old. It does not say anything about sleep training a 3-month-old. It also does not measure whether crying for long stretches during the intervention itself causes acute distress, only whether that distress shows up in measurable outcomes later. Short-term distress is a real cost. The paper is not arguing it is not.

The paper also did not test "no-cry" methods directly because those methods do not have the same level of clinical trial infrastructure behind them. That does not mean they do not work. It means the evidence base for them looks different.

What I tell parents in consults

I do not use unmodified extinction. I have never needed to. The plans I write for clients start with bedtime fading and graduated response, and most families never need anything more aggressive than that. The research tells us extinction works and does not cause harm. It does not tell us it is the only thing that works.

What the research does support, every single time, is consistency. The single biggest predictor of sleep training outcomes across all 16 trials in the analysis was whether parents stuck to the protocol for at least 7 nights. Method mattered less than execution.

If you are sitting on the fence, the research is on your side. Whichever side you choose.