Assessment is a fundamental component of physics and science education; however, the implementation of e-assessment faces unique challenges related to the complexity of abstract concepts and the limitations of comprehensively mapping research trends. This study aims to analyze publication trends, collaboration patterns, knowledge structures, and future development directions of e-assessment in physics and science education through a bibliometric approach. Data were collected from the Scopus database in March 2026 using keywords related to e-assessment and physics or science education, yielding 190 documents from 2016 to 2026. Analysis was conducted using VOSviewer for network mapping and visualization, and Microsoft Excel for preliminary analysis. The results indicate that the number of publications increased significantly following the COVID-19 pandemic, with the United States as the primary contributor and Indonesia ranking fourth. The mapping reveals that research focus has shifted from merely applying digital technology toward integrating artificial intelligence and large language models into adaptive and personalized assessment. New findings also indicate the emergence of atomized problem-solving approaches and the use of translanguaging as an inclusive and equitable assessment practice. In conclusion, e-assessment in physics and science education is evolving toward a convergence of technological innovation and robust pedagogy, with the primary benefit of providing a roadmap for researchers, educators, and policymakers in developing assessments that are adaptive, personalized, and responsive to the cognitive, linguistic, and cultural diversity of students in the digital age.