The National Research Council ([NRC], 2001) has stated that formative assessment is essential to fulfilling the mission of the National Science Education Standards (NRC, 1996). This paper explores the relative utility of four different types of formative assessment prompts in eliciting middle school students' ideas about sinking and floating. Students' written responses and statements in classroom discussions around each of the prompts are compared. Results indicate that the relative success of the prompts in eliciting a range of conceptions may depend on the openness and familiarity of the prompts. Prompts with fewer constraints and unfamiliar settings elicited a range of student conceptions in writing; however, discussions surrounding the prompts were more likely to elicit ideas at the expected level, or no student ideas at all. A comparison of the prompts revealed that the diversity of students' responses in writing was not reflected in classroom discussions. The study suggests that open-format formative assessment prompts may function better when used as a basis for teachers to elicit a range of student ideas in writing, whereas constrained-outcome space prompts may be more appropriate for whole-class conversations that focus students upon scientifically appropriate responses. (Contains 9 tables and 7 figures.)