ObjectivesAssessment of autism‐related behaviours (ARBs) in children has generally been limited to direct observations in clinical settings or informant‐based reports. The widespread availability of video‐streaming devices has made home observations of children's ARBs feasible. This approach could enable assessment of the generalization and durability of interventions and may be able to overcome methodological limitations of predominant current assessment approaches (response biases, limited sensitivity to treatment).Design and MethodsForty‐four autistic children and their families participated in a repeated‐measures study with a correlational design. Approximately 10 hr of unprompted behaviour at home were videorecorded over the course of a week (2 hr per day) for each participant. Gold standard measures of ARBs were also administered (ADOS‐2 and ADI‐R). Two home‐based observational measures of ARBs utilizing streaming video were developed and evaluated: the ARCHER and the CHEERS. Trained independent evaluators made ratings on the ARCHER, CHEERS and an observational measure of parental responsiveness.ResultsCorrelations with the ADOS‐2 and ADI‐R were .47 and .34 for ARCHER scores and .51 and .48 for CHEERS scores, respectively. In linear mixed models, more responsive parenting was associated with fewer ARBs on a daily basis. Children spent their afternoons engaged in many typical activities including electronics, homework and games with family members, and ARBs were more prominent in some of these contexts (e.g., electronics) than others (e.g., family games).ConclusionsHome‐based observational assessment of ARBs may be useful for clinical and descriptive research.