This paper examines the effectiveness of instructional strategies that lower cognitive load by reducing task complexity (intrinsic cognitive load). Three groups of 13-year-old students were required to learn a mathematical task under different conditions. One group (Isolated) followed a strategy that used part-tasks where the constituent elements were isolated from each other (element isolation). A second group (Integrated) received whole tasks where all elements were fully integrated, and a third group (Mixed) followed a mixed strategy progressing from part-tasks to whole-tasks. Results indicated that the part-task strategy was effective in lowering cognitive load for all students, but only benefitted learning for students with low prior knowledge. In contrast, students with a higher prior knowledge learned significantly more having studied whole tasks during instruction compared with part-tasks. The mixed-mode method proved to be ineffective for both levels of prior knowledge. These results are discussed in terms of cognitive load theory. Copyright © 2006 John Wiley & Sons, Ltd.