This study examined whether and why assigning children to a segmented inquiry task makes their investigations more productive. Sixty-one upper elementary-school pupils engaged in a simulation-based inquiry assignment either received a multivariable inquiry task (n = 21), a segmented version of this task that addressed the variables in successive order (n = 21), or could formulate a task themselves (n = 19). Results showed that children are naturally inclined to pose single-variable inquiry questions. Segmented tasks, in addition, invoked more systematic but equally comprehensive investigations than a single, unsegmented task. More systematic experimentation was associated with more valid inferences and beliefs. These findings demonstrate that dividing a multivariable inquiry task into a series of single-variable subtasks facilitates the control of variables rather than the control of the learning process, and promotes inference performance and conceptual understanding. (Contains 2 tables and 1 figure.)