Several empirical studies are discussed which indicate that adapting the level of support, the difficulty of the learning task, or support as well as difficulty has beneficial effects on learning and transfer test performance. Three conclusions can be drawn from the presented studies. First, both the level of available support and the difficulty of the learning task are important dimensions to take into account for dynamic task selection. Second, performance alone is often an insufficient basis for task selection and should be complemented with, for instance, invested mental effort so that mental efficiency drives the task selection. Third, the 4C/ID-model proved to be not only useful to develop fixed training programs but also to develop adaptive forms of instruction.