It is well known that between two thirds and three fourths of children with autism also have intellectual disability. Some of these children were not diagnosed as to have autism due to diagnostic overshadowing between autism and severe forms of intellectual disability.

The objective of this research was to determine the capability of autism index (AI) to correctly identify individuals who have been already diagnosed as having autism.

The sample consisted of 64 participants with moderate, severe and profound intellectual disabilities who have been previously determined to meet DSM-IV-TR criteria for autism spectrum disorder. Autism index was derived by Gilliam Autism Rating Scale (GARS-2).

It was revealed that AI was very high in 39 (60.9%) participants meaning that they were very likely to have autism according to GARS-2. In addition, probability of autism in 18 (28.1%) participants was labeled as “possibly”. AI failed to detect 7 children with autism spectrum disorders.

Sensitivity of AI is quite acceptable (88.9%) providing all children with scores higher than 69 (possibly and very likely as having autism) are considered to have autism spectrum disorder.