An assessment-oriented design-based research model was applied to existing inquiry-oriented multimedia programs in astronomy, biology, and ecology. Building on emerging situative theories of assessment, the model extends prevailing views of formative assessment "for" learning by embedding "discursive" formative assessment more directly into the curriculum. Three twenty-hour curricula were designed and aligned to content standards, and three levels of assessments were developed and used to assess and enhance learning for each curriculum. These assessments included three or four informal "activity-oriented" quizzes and discursive formative feedback rubrics supporting collective discourse, a "curriculum-oriented" examination of individual conceptual understanding, and a "standards-oriented" test measuring aggregated achievement of targeted standards. After two design-research cycles, worthwhile scientific argumentation and statistically significant gains were attained for two of the three packages on the exam and test. Achievement gains were comparable to or larger than those of students in comparison classrooms. Many existing innovations could be enhanced and evaluated in this fashion; designing these strategies directly into innovations could have an even greater impact on discourse, understanding, and achievement. (Contains 3 tables, 6 figures and 8 notes.)