Background Observational tools can help refine practice design and guide the creation of effective learning environments. The intention of this study was to design and validate an observational instrument for assessing physical literacy that remains more faithful to the philosophically complex and holistic nature of the concept. Methods Framed by concepts of ecological dynamics, the emergent games-based assessment tool enables capture of children's interactions with their environment, providing insight on the manifestation of physical literacy within physical education games. The design and validation of the instrument consisted of a multistage process: (1) design of the observational instrument and establishing face validity; (2) pilot observation study; (3) expert qualitative and quantitative review to establish content validity; (4) observation training; and (5), establishing observer reliability. Results Following expert qualitative and quantitative evaluation, Aiken's V coefficient was used to determine content validity. Results achieved demanding levels of validity (V ≥ 0.78) for all retained measurement variables. Cohen's κ values for inter- and intra-observer reliability ranged from 0.331 to 1.00 and 0.552 to 1.00, generally reporting “substantial” agreement during inter-observer analysis and “substantial” to “almost perfect” agreement during intra-observer analysis. Conclusions The final model of the emergent games-based assessment tool, with 9 ecological conceptualisations of behaviour, 15 measurement variables, and 44 categorical observational items was found to be valid and reliable, providing both educators and researchers with a useful mechanism to assess physical literacy during gameplay.