A relaxant treatment given to premature babies reduces their risk of developmental problems, US experts say.
Scientists are looking for ways to reduce complications for babies
The New England Journal of Medicine reports babies given nitric oxide treatment were half as likely to have problems as those not at age two.
There have been concerns over the treatment, which has been linked to an increased risk of brain haemorrhage.
A second study suggests this means the smallest babies should not be given the treatment.
Premature babies with under-developed lungs are at risk of brain injury and developmental problems.
Nitric oxide is known to relax blood vessels in the lungs and to release growth factors for more blood vessels.
Helping a baby's lungs work more effectively would mean more oxygen was available in their bodies, and problems linked to a lack of oxygen should be reduced.
Nitric oxide is already used to treat full term babies who have chronic respiratory distress.
Stanford University scientists have been following a group of 138 toddlers since birth.
They had already found adding tiny amounts of nitric oxide treatment to the oxygen given to babies reduced the risk of chronic lung disease, severe bleeding into the brain and death in premature infants with respiratory distress syndrome.
Neurologists have now gone back to the same group and compared the development at the age of two of the 68 children given nitric oxide with the 70 who were not.
They found 24% of the children who received nitric oxide as babies had delayed mental development or a disability, such as blindness, cerebral palsy or hearing loss, compared to 46% of those who received standard treatment of oxygen with no nitric oxide.
Sixteen per cent who received nitric oxide treatment were seen to have delayed brain development compared to 34% of those who did not.
Michael Schreiber, professor of paediatrics at the University of Chicago, said: "This is the first therapy for premature infants that has demonstrated a significant impact on brain development.
He added: "Despite everything we have tried in neonatology, no therapy has ever improved cognitive function.
"Babies born at two pounds have only a 50% chance at two years of age of being considered totally normal - meaning no cerebral palsy and normal IQs."
A second study in the New England Journal of Medicine, by a team from Stanford University School of Medicine, looked at 400 babies born very prematurely.
All the babies they assessed were born before 34 weeks of gestation and weighed between 14 ounces (401 grams) to 3.3 pounds (1,500 grams)
The babies were also all suffering from respiratory failure despite being given other treatments.
It was found that nitric oxide was linked to a higher rate of brain haemorrhage in the most premature babies - 2.2 pounds (1,000 grams) or less.
But in bigger babies, it reduced the risk of lung damage to 50%, from 69% in the dummy treatment group.
Krisa van Meurs, the neonatologist who led the research, said: "This study suggests that the key to the effective use of inhaled nitric oxide may lie in choosing the right patients."
But she added: "Because of its effects on bleeding, it may not be useful in certain critically ill babies."
David Field, professor of neonatology at the University of Leicester, said the US findings did not give the green light for using nitric oxide treatment, but were useful pointers to further research.
"The babies in the US study who were not given nitric oxide did not do as well as babies in the UK would generally do."
But he said: "They did find a difference. The difference in the UK might not be as great, but it could help."