One of the most important contributory factors of software correctness, hence reliability, is the application of Formal Methods. These methods should be widely used in practice, but their acceptance by industry is well below what it should be. The most commonly mentioned way of improving on this situation is to reform education. This must address two issues. The first is to establish an appropriate concept of Formal Methods, one which is both sound and practicable. The second is to establish the proper way of teaching the appropriate methods. This paper addresses these two issues. First, it proposes a concept of Formal Methods and examines various misconceptions about these methods. The main tenet of the paper is that Formal Methods should give priority to the formalisation of specifications of system components over the production of completely formal proofs. Second, the paper reports on the author’s experience in teaching Formal Methods and related topics at introductory (MSc conversion) level over many years. Examples of simple specifications are given, their contribution to students’ education discussed, and a number of teaching recommendations are proposed.