Naturalistic developmental behavioural interventions (NDBI) are a group of evidence-based early interventions for autism spectrum disorders (ASD). Caregivers can be trained to deliver NDBI strategies during interactions with their young child with ASD. However, NDBI research predominantly comes from high-income countries, and the evidence base for NDBI in low/middle-income countries (LMICs) and across cultures is lacking. To understand the ‘fit' of an NDBI approach in LMICs, it is crucial to understand and be able to measure the nature of interactions between caregivers and their children with ASD and the context wherein caregiver-child interactions occur. This study sought a) to evaluate the utility of a specific measure of caregiver-child dyadic interactions and b) to examine daily routines in which caregiver-child interactions occurred in low-resource South African contexts. Methodology Children with ASD (between 18-72 months old) and their ≥18-year-old caregivers were recruited under a larger project. Interactions of 21 caregiver-child dyads were video-recorded using a standardised parent-child interaction (PCI) protocol with two 6-minute-long free-play sessions (Part I: child explored the room and available toys while the caregiver remained seated; Part II: caregiver interacted with their child as they would at home). Two research-reliable raters rated the videos using 16 items from the Joint Engagement Rating Inventory (JERI), a 7-point Likert scale behavioural coding system. Reliability and descriptive analyses were conducted. Structured interviews were conducted with ten caregivers using the Parent Survey of Home and Family Experiences (PSHFE) to explore the context of daily routines. Descriptive analyses were performed. Results For caregiver-child interactions, observer agreement for 12 of 16 items was reasonable, with weighted kappas (within 1 scale point) of 0.66-1, an estimated accuracy of 88-99%, and percentage agreements of 75-100% for all items. Ratings for items across Parts I and II of the JERI showed variability without any ceiling effects. Six items showed floor effects. Most caregiver item ratings were at the mid-point of the 7-point Likert scale. In Part II, children used more expressive language and paid more attention to their caregivers. On the PSHFE, most children participated daily in various child routines, play and early literacy activities with mothers as main partners. Most children never participated in spiritual and community activities, typically due to the child's age, safety and other reasons not specified in interview response categories. Conclusion Reliability, floor/ceiling, behavioural and Part I vs Part II profiles suggested that the JERI, used for the first time in a South African context, has potential utility both to describe caregiver-child interactions and be used as an intervention outcome measure in LMICs. The PSHFE results provided contextual data of common daily activities into which NDBI strategies could be embedded to support child generalisation of skills in South Africa.