Generalizability and decision studies provide a mathematical framework for quantifying the stability of a given number of measurements. This approach is especially relevant to the task of obtaining a representative measure of communicative behavior in young children and supports an alternative to the debate regarding which type of assessment yields the most representative scores. The current article provides a report of a generalizability and decision study on 63 toddlers with developmental delay who were treated for 6 months using an intervention that targeted communication and vocabulary goals. Two variables--rate of intentional communication acts and rate of different words--were measured across three assessment contexts at four communication sampling periods. Results verified that measurement stability increased with time and development for both variables, regardless of the type of assessment procedure used.