BBC News X-rays fail to spot up to nearly a third of fractures in the bones of the hip and pelvis, research suggests.
Less obvious fractures may be missed on normal X-rays
US doctors from Duke University warn that relying on standard X-rays alone to give a clean bill of health could lead to wrong diagnoses and law suits.
When they re-checked the X-ray results of 92 patients using a more detailed scanner, called MRI, they found 35 fractures that had been missed.
The findings are published in the American Journal of Roentgenology.
Dr Charles Spritzer, who led the research, said: "The diagnosis of traumatic fracture most often begins and ends with X-rays of the hip, pelvis, or both.
"In some cases though, the exclusion of a traumatic fracture is difficult."
In the study of patients complaining of pain after an injury to the hip or pelvis, 13 with normal X-ray findings were found to have a collective total of 23 fractures seen on MRI (magnetic resonance imaging).
Another 15 patients with abnormal X-rays had 12 additional pelvic fractures seen on MRI that otherwise would not have been identified.
Ultimately, it comes down to clinical acumen
Dr Tony Nicholson of the Royal College of Radiologists
And in 11 patients, MRI showed no fracture after X-rays had suggested there might be one.
The researchers say it is worth using MRI as well as an X-ray if doctors are in any doubt, particularly since hip patients tend to be frail and elderly with a higher than average risk of complications and death.
Dr Spritzer said: "Accurate diagnosis of hip and pelvic fractures in the emergency department can speed patients to surgical management, if needed, and reduce the rate of hospital admissions among patients who do not have fractures.
"Use of MRI in patients with a strong clinical suspicion of traumatic injury but unimpressive X-rays has a substantial advantage in the detection of pelvic and hip fractures."
Dr Tony Nicholson, from the Royal College of Radiologists, said the findings quantified something already known or suspected.
But he did not think it would be feasible or sensible to give every patient an MRI scan.
"Ultimately, it comes down to clinical acumen. If an elderly patient has persistent pain even though their X-ray shows only minor arthritis, an MRI would be a very reasonable request to check that there is not a fracture.
"It is always worrying when something is not diagnosed immediately, but I think most clinicians are first class and would follow up a patient appropriately."