With reference to existing literature and experiences gained from both domestic and international large-scale assessments in science, this study aims at developing an assessment framework and some illustrative test items, which will provide guidelines and concrete examples for item and test developments, and the administration of a national survey in subsequent studies. The assessment framework, designed for the 4(superscript th), 6(superscript th), 9(superscript th), and 12(superscript th) graders, consists of, first of all, a scheme to categorize students' major learning outcomes in content domains including biology, chemistry, physics and earth science. Within these subject areas, students' expected performances in terms of their proficiency in science practices, including understanding and applications of science concepts and principles, scientific enquiry skills, quantitative reasoning skills, nature of science, and attitudes and other affective factors toward learning science are specified. In order to follow up students' learning outcomes on a long term basis, numerous examples of students' expected performances, varying from basic, proficient to advanced levels are developed. Serving as guidelines for subsequent studies along this line, some general principles of quality assessment and specifications for test development in science, as well as illustrative test items, are developed in this study. The study was coordinated by the first author, while the other coauthors respectively organized research teams consisting of science teachers, scientists, and science educators of different content areas in developing the assessment framework and illustrative test items. The work done in this study are of a preliminary nature, field tests and further refinement may be necessary.