Technological interventions for teaching children with autism spectrum disorders (ASD) are becoming popular due to their potentials for sustaining the attention of children with rich multimedia and repetitive functionalities. The degree of attentiveness to these technological interventions differs from one child to another due to variability in the spectrum. Therefore, an objective approach, as opposed to the subjective type of attention assessment, becomes essential for automatically monitoring attention in order to design and develop adaptive learning tools, as well as to support caregivers to evaluate learning tools. The analysis of facial expressions recently emerged as an objective method of measuring attention and participation levels of typical learners. However, few studies have examined facial expressions of children with ASD during an attention task. Thus, this study aims to evaluate existing facial expression parameters developed by “affectiva”, a commercial engagement level measuring tool. We conducted fifteen experimental scenarios of 5Â min each with 4 children with ASD and 4 typically developing children with an average age of 8.8 years, A desktop virtual reality-continuous performance task (VR-CPT) as attention stimuli and a webcam were used to stream real-time facial expressions. All the participants scored above average in the VR-CPT and the performance of the TD group was better than that of ASD. While 3 out of 10 facial expressions were prominent in the two groups, ASD group showed addition facial expression. Our findings showed that facial expression could serve as a biomarker for measuring attention differentiating the groups.