I think that a lot of children are diagnosed as autistic who would not have been in the past. This occurs at both ends of the scale: children who might in the past have been called 'naughty' or 'odd' or 'maladjusted' or 'emotionally disturbed' are now diagnosed with 'high-functioning autism/ Asperger syndrome', while others who would have been called 'mentally retarded' without further classification are now diagnosed as 'autistic' if they have particular problems with communication and interaction.
This is, I suspect partly because of changes in the educational system, both good and bad. In the past, many children with low IQs were simply dismissed as 'ineducable', whereas now they are seen as needing education, and their problems need to be analyzed more carefully, making an 'autism' diagnosis more likely. And children of higher IQ who don't fit into the school system may be more likely to have their specific needs analyzed, rather than just being ignored or treated as 'discipline problems'. These are good changes. On the negative side, pressures to get schools to 'reach targets' may lead to greater pressures to get a semi-medical diagnosis for children who are not doing all the things demanded by policymakers and governments. OFSTED or NCLB may only be prepared to reduce demands to force a square peg into a round hole if a definite diagnosis is given.
I also think that there is some increase in incidence of autism, due to the fact that extremely premature babies and those with other life-threatening perinatal problems are more likely to be autistic, and a far higher proportion of babies with such problems survive than in the past - 30 years ago many such babies never lived to be diagnosed as autistic. There may also be some environmental factor that is increasing the risk of autism - if there is, then I suspect that it is one that exerts its main influence prenatally; and I think that there needs to be more research on factors that may affect pregnant women and their unborn babies. But I don't think that there is the massive increase in autism that some people imply; and this is a view held by many researchers on autism.
ETA: Here is a link to Helen Heussler's letter to the British Medical Journal in 2001, which gives some data to suggest that diagnostic criteria have indeed changed since the 1970s, and that this could have contributed to the apparent rise in incidence.
http://www.bmj.com/cgi/content/full/323/7313/633?ijkey=b36b9709203810ef3ef802a7009018cfcec0797a&keytype2=tf_ipsecsha