Purpose Narrative skill, a child's ability to create a temporally sequenced account of an experience or event, is considered an important domain of children's language development. Narrative skill is strongly predictive of later language and literacy and is emphasized in curricula and educational standards. However, the need to transcribe a child's narrative and the lack of psychometrically justified scoring methods have precluded broad consideration of narrative skill among practitioners. We describe the development and validation of the Narrative Assessment Protocol-2 (NAP-2), an assessment of narrative skill for children ages 3-6 years, which uses event-based frequency scoring directly from a video recording of a child's narrative. Method The NAP-2 underwent a rigorous development process involving creation of four wordless picture books and associated scripts and identification of a broad item pool, including aspects of narrative microstructure and macrostructure. We collected two narratives from each of 470 children using the NAP-2 elicitation materials and scored each with the 60 items in the initial item pool. Results Cross-validated exploratory factor analyses indicated a single narrative skill factor. Rasch measurement analysis led to selection of 20 items that maintained high reliability while having good fit to the model and no evidence of differential item functioning across books and gender. Conclusions The NAP-2 offers a psychometrically sound and easy-to-use assessment of narrative skill for children ages 3-6 years. The NAP-2 is available freely online for use by speech-language pathologists, educational practitioners, and researchers. Supplemental Material https://doi.org/10.23641/asha.11800779.