This paper describes two empirical studies that informed the development of a Task Wizard, a computer-based system for presenting task design guidance. Two empirical studies compared specialist task designers with experienced teachers using task design observation and task sorting paradigms. In Study 1, designers worked initially in depth and then worked in breadth before returning to depth-first development, unlike teachers who moved from working in breadth initially to working in depth. In Study 2, designers' classifications of tasks were less dominated by deep conceptual topic-based categories than were those of teachers. These results are the reverse of expert/novice differences identified in the literature and have important implications for theories of design expertise in particular, and complex problemsolving and expertise in general. The studies also provide data on task classifications, evaluative criteria, examples of good and bad designs, and design strategies and practices that informed the development of task design guides. The guides are delivered in the form of a Task Wizard, a computer-based platform for supporting production, adaptation and selection of tasks for classroom activities and examination questions.