The study examined children's understanding of scientific inquiry, through the lens of their conceptualization of uncertainty in investigations they had designed and implemented with a partner. These largely student-regulated investigations followed a unit about animal behavior that emphasized the scaffolding of independent inquiry. Participants were children in a 2nd grade class (n = 21) and a combined 4th-5th grade class (n = 31). The primary database consisted of videotaped structured interviews with each pair following completion of their investigation. These interviews were used to analyze whether the child had conceptualized an uncertainty in their study and, if so, the sphere of the uncertainty. This top-level analysis resulted in a taxonomy of 5 spheres of uncertainty: (a) how to produce the desired outcome as uncertain; (b) data as uncertain; (c) trend identified in the data as uncertain; (d) generalizability of this trend as uncertain; and (e) the theory that best accounts for the trend as uncertain. Seventy-one percent of the 2nd graders and 87% of the 4th-5th graders conceptualized 1 or more spheres of uncertainty in their respective research study. Among those who had conceptualized 1 or more spheres of uncertainty, 80% of the 2nd graders and 97% of the 4th-5th graders posited a strategy for how to modify their study to address the uncertainty. Finer grained analysis revealed that in almost every instance the grounds reflected some valid concern and the strategy the child proposed would at least begin to address the identified problem. These analyses indicate that most children developed a rich understanding of how uncertainty enters into scientific inquiry. In this context of reflection on their own scientific inquiry, these children manifested at least a rudimentary understanding of the complex relation between the natural world and knowledge thereof that transcended a naive realism, and manifested aspects of a "knowledge problematic" epistemological perspective.