Elementary school children are capable of reproducing sophisticated science process skills such as observing, designing experiments, collecting data, and evaluating evidence. An understanding of the nature of scientific knowledge requires more than teaching and learning the performance of these skills. It also requires an appreciation of how these actions lead to knowledge generation and shape its durable and tentative nature. Our understanding of activities that support the teaching and learning of the nature of scientific knowledge is still growing. This study compares how scientific knowledge is generated in science laboratories and in elementary school classrooms. Discourse analysis, conducted through an activity theoretical perspective, of conversations between biomedical researchers revealed that nearly every aspect of discussions among scientists concerned uncertainty--establishing what was known with confidence and what remained unresolved. Analysis of conversations between teachers and students in an urban elementary school showed that a similar pattern of identifying and resolving uncertainty was evident when students were conducting independent investigations. All participants identified new knowledge through processes of reflexivity concerned with what was observed, conducted, and interpreted in investigations. Implications for science education are discussed. (Contains 1 table and 2 footnotes.)