Effective instructional explanations help the students to construct coherent mental representations. To do so, one condition is that they must be tailored to students' needs. It is hypothesized that explanations are more helpful if they also explicitly aid the students to detect problems in their mental representations, as this provokes an impasse that motivates students to process the explanation deeply. Participants were provided with a computer-based material on plate tectonics and then with explanatory support in the form of either a tailored explanation preceded by an impasse-trigger (I + E group) or an identical explanation without the impasse-trigger (noI + E group). After the reading of the materials they solved retention and transfer tests; their flawed ideas were also counted. Participants in the I + E group recalled more correct information, generated more transfer solutions, and showed fewer flawed ideas than those in the noI + E group. This indicates that tailored explanations combined with impasse-triggers that make explicit conflicts between the text model and the students' models can indeed foster deep learning.