Guidelines for children's heart and breathing rates are inaccurate and not based on evidence, according to researchers.
These measurements are often used to decide treatment, but the study in the Lancet says current rules would mean half of healthy 10-year-olds were diagnosed with abnormal vital signs.
The study analysed data from 143,346 children.
Experts say the work could help clinicians.
How rapidly a child is breathing and the speed of their heart are important tools for diagnosing illness.
These measurements are compared with charts which show normal and unhealthy levels.
However, the researchers say the healthy ranges have been passed down from one generation of doctors to the next, but have not been based on evidence.
Inaccurate
Their analysis looked at 69 studies on the vital signs of nearly 150,000 healthy children from around the world.
One of the problems in the field is that health levels shift during childhood.
The study showed that on average, heart rates were 145 beats per min in one-month-olds, while it was 113 beats per min in two-year-olds.
The authors produced new charts showing the healthy vital signs at different ages and compared these with the current guidelines.
Dr Matthew Thompson, from the University of Oxford, said: Children with fast heart rates are not being picked up and some are being diagnosed with a fast rate when they're normal.
Our findings suggest that current consensus-based reference ranges should be updated with new thresholds on the basis of our proposed centile charts, especially for those age groups where there are large differences between current ranges and our centile charts, indicating that many children are likely to be misclassified.
Rosalind Smyth, from the University of Liverpool and Alder Hey Children's Hospital, said this study should lead to more research to determine the barrier between normal and abnormal vital signs.
She said: These studies will lead to revised algorithms, risk scores, and guidelines, which will incorporate these limits.
Such instruments will then need to be extensively validated in different settings and populations before they can be incorporated into clinical practice.
