Purpose: In this validation study, we examined the factor structure of the mediated learning observation (MLO) used during the teaching phase of dynamic assessment. As an indicator of validity, we evaluated whether the MLO factor structure was consistent across children with and without developmental language disorder (DLD). Method: Two hundred twenty-four children (188 typically developing and 36 DLD) from kindergarten to second grade completed a 30-min individual mediated learning session on narrative production. Performance during the session was rated using the 12-item MLO by clinicians on affect, behavior, arousal, and elaboration. Exploratory and confirmatory factor analyses were conducted to establish the factor structure and reliability of the MLO. Results: Factor analysis of the MLO suggested a stable three-factor model with adequate fit indices across kindergarten and school-age samples, across both typically developing and DLD subgroups with good to excellent reliability. The final 11-item MLO (one item was removed due to low factor loading) comprises three subscales including (a) cognitive factor, (b) learning anticipation, and (c) learning engagement. Conclusions: The MLO is a valid and reliable instrument for assessing language learning skills in children with and without DLD during dynamic assessment. Practical implications and suggestions for future research addressing the utilization of MLO in dynamic assessment are provided.