Two studies were conducted to test the hypothesis that young autistic children are selectively impaired on emotion perception tasks. Results supporting the hypothesis were found on two of the four measures when the controls used were matched on non-verbal mental age; performance on the other tasks was consistent with global deficits across affective and non-affective domains, rather than specific deficits in emotion perception. When the autistic group was compared with controls matched on verbal mental age, no group differences were found. These results suggest that emotion perception impairment is not likely to be the primary underlying deficit in autism. Additional areas for further investigation were suggested.