When Leo Kanner first described autism in 1943 it was thought to be extremely rare, affecting no more than one to three children in 10,000. Yet it is now so common, with a prevalence of between one in 300 and one in 100, that most people know an individual or family touched by it.

These statistics have given currency to the idea that there is an autism epidemic - and to fears that an aspect of modern life has made children more susceptible. This has fed unfounded scares about the role of vaccines.

Although autism is certainly diagnosed more often than it once was, that does not mean that more children are developing it. As a remarkable book being published next week explains, the condition is probably no more prevalent than it has always been. What has changed is that it is being recognised properly for the first time.

Roy Richard Grinker is the father of Isabel, an autistic girl, now 14. His book Unstrange Minds is a powerful memoir of his family's experience;of the struggle to obtain a diagnosis for Isabel and to have her needs catered for by the education system. It also describes her difficult but warm and rewarding personality and her family's joy as she learns to make sense of a confusing and often frightening world.

It is impossible not to be moved, not least by Isabel's transforming fascination with Monet's water lilies, and the moment when an argument with her non-autistic sister shows that she is starting to understand other people's emotions.

Grinker is also an anthropologist, who has used his skills to examine how cultural attitudes to autism affect our experience of the condition. His compelling argument is that there is no autism epidemic. The apparently rising incidence is the result of broader, earlier and more accurate diagnosis and greater awareness among doctors and patients.

Autism did not feature in the standard psychiatry textbook, the Diagnostic and Statistical Manual of Mental Disorders, until 1980. Later revisions broadened the criteria, placing more children in the autistic spectrum. One even contained an error that inadvertently widened the definition.

These changes, Grinker shows, matched the curve of the supposed epidemic. Children who would once have been considered mentally retarded or schizophrenic are today classed, more correctly, as autistic. So, too, are some higher-functioning children with autism-spectrum disorders, who previously escaped psychiatric attention altogether. These people were always there but have now been given an appropriate label.

Altered parental perceptions have made a difference, too. As autism has been diagnosed more often, it has become less frightening and more socially acceptable. Indeed, much less stigma is attached to this disorder than to mental retardation: many parents of atypical children now want them to be labelled autistic. Grinker's assessment of autism in non-Western cultures adds to his case that complex cultural factors cause the same symptoms to be interpreted differently.

The availability of social, educational and medical services can also matter. Grinker's home state of Maryland, for example, waives many medical fees for autistic children but not for the mentally retarded, creating an incentive for parents and their doctors to lean one way and not the other.

If Grinker is right, the so-called autism epidemic is not to be feared but to be celebrated. It means that children and adults who for centuries have been given the wrong diagnosis, or been missed altogether, are finally being assessed appropriately. That can only be good for research into the condition's origins and treatment. Most importantly, it will help the rest of us to understand and meet autistic people's needs.

Mark Henderson is Science Editor of The Times

Unstrange Minds (Icon Books, £14.99)

Explore Health

• Expert Advice

• Health Features

• Mental Health

• Alternative Medicine

• Child Health

• Health Club