Background: It is unclear whether the increase in autism prevalence over the past two decades is a real increase or due to changes in diagnosis and ascertainment of autism spectrum disorders (ASD). Objectives: The aim of this study was to examine the trend of ASD over time and the possible effects of diagnostic substitution, changes in diagnostic criteria and eligibility for service provision. Methods: A population based study was conducted among a birth cohort of children born in Western Australia between 1983 and 1999 and diagnosed with an ASD up to 2004. The trend in the birth prevalence of ASD and corresponding diagnosis of intellectual disability (ID) during the study period was investigated. Sub-group analysis of children aged five years and under where complete data were available was also conducted to examine the incidence and trend of ASD diagnoses over time. Results: Prevalence of all ASD increased by 29.6% per annum, with autism rising by 20.8% each year. Up to 1992, diagnosis of ID was more common but it was then surpassed by ASD and with the prevalence of severe ID decreasing by 6.2% each year. From 1992, for children aged five years or less the incidence of autism increased significantly with an annual increase of 27.4% as did PDD-NOS with an annual increase of 35.1%. In the corresponding years the incidence of severe ID fell annually by 6% and mild-moderate ID continued to increase by 6.6% per annum. Conclusions: The rise in birth prevalence of ASD and autism by birth year in WA can be partly, but not fully, explained by a diagnostic substitution of ID to autism. Introduction of ASD assessment in WA and iincreased government funding and availability of services at the time of the increase in ASD is likely to have contributed to the rise.