A multitude of research on facial emotion recognition (FER) in Autism Spectrum Disorders (ASD) have been published since several years. However, these studies have mainly used static high intensity stimuli, including adult and/or children facial emotions. This current study investigated FER in children with ASD using an innovative task, composed of a combination of static (114 pictures) and dynamic (36 videos) subtests, including children, adolescent, and adult male and female faces, with high, medium, and low intensity of basic facial emotions, and neutral expression. The ASD group consisted of 45 Tunisian verbal children, and the control group consisted of 117 tunisian typically developing children. Both groups were aged 7–12 years. After adjusting for sex, age, mental age, and school grade, the ASD group scored lower than controls on all tests except for the recognition of happiness and fear in the static subtest, and the recognition of happiness, fear, and sadness in the dynamic subtest (p ≥ 0.05). In the ASD group, the total score of both the static and the dynamic subtest were positively correlated with the school grade (p < 0.001), but not with age, or mental age. Children with ASD performed better in recognizing facial emotions in children than in adults and adolescents on videos and photos (p < 0.001). Impairments in FER would have negative impact on the child's social development. Thus, the creation of new intervention instruments aiming to improve emotion recognition strategies at an early stage to individuals with ASD seems fundamental.