Purpose: This research investigated the use of a dynamic assessment (DA) to identify differences among young children with severe disabilities, which would predict progress in learning behaviors indicating coordinated joint attention (CJA). Method: Six children 10-24 months of age were enrolled in a 16-week treatment for behaviors indicating CJA, specifically triadic gaze (TG), which is a 3-point gaze shift between object and adult. An initial static assessment documented the children's eligibility for the study and their baseline performance of TG. DA procedures were then implemented to determine each child's performance with examiner support in producing behaviors suggesting joint attention (i.e., tracking, gaze toward an object or an adult, scanning between objects, scanning an object and adult, and TG). Results: Results demonstrated differences among children during the DA via a DA score and a behavioral profile. These results were predictive of differences among children in subsequent learning of TG. Conclusion: These data support the validity of DA for describing heterogeneity among young children with severe disabilities who look similar on static assessment but appear differentially ready to learn behaviors associated with joint attention. This knowledge will assist clinicians in planning more efficacious services for young children who struggle to communicate and are at risk for extended therapeutic needs. (Contains 3 tables and 1 figure.)