Examples are often recognized as important in teaching conceptual material, yet little research has been done concerning the best use of examples in attempts to remediate misconceptions. This study questions the effectiveness of a traditional teaching-by-example technique. Results indicate that when students hold a misconception, presenting a principle with supporting examples to show the range of application of the principle may be ineffective. Rather, it appears that examples are more effective when they help students draw on and analogically extend existing valid physical intuitions in constructing a new conceptual model of a target situation. To help students in this constructive effort, first, the examples used must be understandable and believable to the students, not simply to the teacher or textbook author. Second, even when an example is compelling to the student, it may not be seen as analogous to target problems drawing out a misconception. In that case, analogy relations may need to be explicitly developed. Third, qualitative, visualizable models may need to be developed which give mechanistic explanations for phenomena.