Greater awareness, and people being diagnosed as autistic who in the past may have had other diagnoses or none. People with relatively mild problems might have been classed as 'emotionally disturbed', 'maladjusted', 'language delayed' or just 'odd', while those with more severe problems might have been classed as 'mentally handicapped', 'childhood schizophrenic' (a diagnosis almost never used nowadays) or 'psychotic'.
A study by Brugha et al (2011) involved assessing several thousand unselected people according to the current criteria for autistic spectrum disorder. The rate was about 1 per cent and it was the same for children, younger adults and older adults, suggesting that actual rates of autism had not changed, though diagnosis rates had. (Admittedly this was a few years ago, but the biggest increase in autism diagnosis rates occurred during the 90s.)
There may be a slight increase in actual rates, due to very premature or sick babies surviving at increased risk of autism, where 40 years ago, they would likely have died. Also, more speculatively, it may be that increased average (of fathers as well as mothers) of having children may somewhat increase the chances of autism.
But the main causes of autism are genetic. And if 4 out of 5 children in a family are autistic, this indicates a strong genetic tendency in that family.